NASA CR-134610 SRD-74-034 STUDY, SELECTION, AND PREPARTION OF SOLID CATIONIC CONDUCTORS by W. L. Roth and O. Muller GENERAL ELECTRIC COMPANY prepared for NATIONAL AERONAUTICS AND SPACE ADMINISTRATION NASA Lewis Research Center Contract NAS 3-17339 https://ntrs.nasa.gov/search.jsp?R=19740018385 2020-03-23T09:02:52+00:00Z 1. Report No. NASA CR-134610 2. Government Accession No. 4. Title and Subtitle STUDY, SELECTION, AND PREPARATION OF SOLID CATIONIC CONDUCTORS 7. Author(s) W. L. Roth and O. Muller 9. Performing Organization Name and Address General Electric Company P. O. Box 8 Schenectady, New York 12301 12. Sponsoring Agency Name and Address National Aeronautics and Space Washington, D. C. 20546 Administration 3. Recipient's Catalog No. 5. Report Date April 1974 6. Performing Organization Code 8. Performing Organization Report No. SRD-74-034 10. Work Unit No. 11. Contract or Grant No. NAS3-17339 13. Type of Report and Period Covered Final Report 11/13/72 - ll/ 12/ 73 14. Sponsoring Agency Code 15. Supplementary Notes Project Managers: J. B. Bush, General Electric Corporate Research and Development, P. O. Box 8, Schenectady, N. Y. 12301; Albert C. Antoine, Energy Conversion and Material Science Division, NASA Lewis Research Center, Cleveland, Ohio 16. Abstract Crystal chemical principles and transport theory have been used to predict structures and specific compounds which might find application as solid electrolytes in rechargeable high energy and high power density batteries operating at temperatures less than 200°C. Structures with 1-, 2-, and 3 -dimensional channels were synthesized and screened by nuclear magnetic resonance, di- electric loss, and conductivity. There is significant conductivity at room temperature in some of the materials but none attain a level that is comparable to beta-alumina. Microwave and fast pulse methods were developed to measure conductivity in powders and in small crystals. 17. Key Words (Suggested by Author(s)) Ionic conductors Solid electrolytes Batteries Nuclear Magnetic Resonance Transport numbers 19. Security Classif. (of this report) Unclassified 18. Distribution Statement Unclassified, Unlimited 20. Security Classif. (of this page) Unclassified 21. No. of Pages 22. Price* 69 *For sale'bythe National Technical Information Service, Springfield, Virginia 22151 NASA-C-168 (Rev. 6-71) TABLE OF CONTENTS I. SUMMARY 1 II. INTRODUCTION - 2 A. Background 2 B. Technical Approach and Experimental Methods 3 1. Structure Selection 3 2. Synthesis and Fabrication of Materials 5 3. NMR Screening 6 4. Conductivity and Transport Measurements 6 a. Two-phase measurement of conductivity 6 b. Microwave measurement of conductivity of powders - - 7 c. Conductivity, transport, and pulse measurements on micro-specimen 9 HI. VALIDATING EXPERIMENTS 14 A. Bridge Measurements 14 B. Applications of Pulse Technique to Beta-Alumina 14 IV. RESULTS AND DISCUSSION OF RESULTS 17 A. LiNb6O15F~Type Compounds 17 B. Hollandite-Type Compounds 22 C. Cordierite-Type Compounds 27 D. NaNbO2F2-Type Compounds 31 E. Phases with a Defect Perovskite Structure 35 F. Sodium Tungsten Bronze-Type Compounds 40 G. NaLiGe^g-Type Compounds 41 H. Li2Ge7O15-Type Compounds 46 I. Hexagonal NaLnF4-Type Compounds 54 J. Dalyite-Type Compounds 60 V. CONCLUSIONS AND RECOMMENDATIONS- - 63 A. Conclusions 65 B. Recommendations > • 65 VI. REFERENCES- - 67 STUDY, SELECTION, AND PREPARATION OF SOLID CATIONIC CONDUCTORS W. L. Roth and O. Muller I. SUMMARY A proposal entitled " Study, Selection, and Preparation of Solid Cationic Conductors" was submitted to NASA in May 1971, in response to RFP No. 416187. It described a research program aimed at the identification and preparation of materials for use as nonporous solid ionic conductors in re- chargeable high-energy and power density batteries operating at temperatures less than 200°C. The only materials capable of meeting the required objectives that are now identified are members of the beta-alumina family. A prelimi- nary search for such solid electrolytes was carried out between July 1971 and June 1972 under NASA Contract NAS3-15692. The work was summarized in Report No. CR-120964, dated September 1972. More than twenty compounds were screened, and encouraging results were obtained for nine materials. None were proved to be good ionic conductors at low temperature, in large part because dense single-phase ceramics were not available for measure- ment of ion transport and conductivity. This report summarizes a continuation of the search which was carried out between December 1972 and November 1973 under NASA Contract NAS3-17339. The structural criteria for predicting materials with high ion mobility at ambient temperature had been validated in the first phase of the study. These criteria were used to predict new candidates with particular emphasis on materials which could be fabricated in practical systems. The major modification in the technical approach was in the area of evaluation of materials. Two new techniques for measuring ionic conductivity and trans- port were developed and validated. By use of these methods it now is possible to evaluate candidate materials in the form of powders, small crystals, or fragments of ceramics or glasses. More than twenty-eight materials dis- tributed among eight structure types were synthesized and evaluated. At least nine of these had conductivities of order 10~6 ohm'1 cm'1 at room temperature, but additional work would be required to establish the level of ion transport. The principal conclusions of the study are as follows: 1. Satisfactory crystal chemical and measurement tools for the pre- diction and the evaluation of solid ionic conductors have been developed and validated, 2. It is unlikely that the required high levels of ion transport can be sustained in one-dimensional tunnel structures, and 3. There are reasonable prospects for developing high ionic conductivity at moderate temperature in three- and two-dimensional interconnected channel structures. Specific recommendations are made for future work on a broad family of compounds which appear to satisfy the principal requirements for use in a low-temperature, high-energy and power density battery. II. INTRODUCTION A. Background The objective of this research program is the identification and prepara- tion of materials for useasnonporous solid ionic conductors in rechargeable high-energy and power density battery systems operating at temperatures less than 200°C. The primary requirements for the electrolyte are that: 1. The ionic conductivity at operating temperature must be sufficiently high to allow high power density. In practice this means the conductivity should be at least 10~3 ohm'1 cm"1 at 200°C. 2. The cation transport number should be unity, i.e., the current carried by anions and electrons must be negligible. 3. The electrolyte must conduct ions of reactants with high oxidizing and reducing power to achieve high-power density. This requirement eliminates silver conductors and focuses attention on cations in the first and second columns of the periodic table. 4. It must be chemically stable with respect to the cell reactants and products. 5. It must be possible to fabricate the material into thin separators which are mechanically rigid and of low porosity to prevent mixing of the cell reactants. The only solids known to satisfy these requirements are the beta- alumina sodium ion conductors. The work to be described in this report is a continuation of a one-year program carried out under National Aeronautics and Space Administration Contract NAS3-15692(2) with the objective to conceive, synthesize, and characterize new solid ionic conductors. In that program a large number of materials were selected on the basis of thermodynamic and crystal-chemical criteria and were subjected to preliminary synthesis and evaluation of their potential as solid ionic conductors. A novel technique, nuclear magnetic resonance (NMR), was employed for the preliminary evaluation. This technique permits the rapid screening of loose powders, thus eliminating the need for the time-consuming preparation of dense, sintered compacts of a large number of materials. Among the materials subjected to NMR analysis, many showed encouraging evidence for ionic motion. These materials are Li3AlN2, Li3BN2, LixTa2O5, LixV2O5, NaPrF4, NaYbF4, LiNb6O15F, and LiNO3 • Z which is LiNOs occluded in the pores of zeolite 4A. Dense samples of the vanadates and tantalates were prepared and ionic conductivity measurements were carried out. The vanadates were shown to be partially ionic and partially electronic conductors with the level of electronic conductivity strongly dependent on the stoichiometry. B. Technical Approach and Experimental Methods The search for new low-temperature cation conductors consists of several interrelated tasks. These are: the application of crystal chemical criteria to predict the structures of materials in which high cation mobility at low temperature is theoretically possible, the synthesis and fabrication of selected candidates, and the characterization of conductivity and transport in synthesized candidates. Several conclusions that were reached as a result of the work under Contract NAS3-15692 led to refinements and modifications in the technical approach followed for the work under this contract. One was a decision to reduce the time-consuming work which is required to fabricate dense ceramics until there is evidence the material is a good conductor of ions. A second was to prepare candidates as pure single-phase specimens in order to avoid the ambiguities inherent in the interpretation of conductivity and transport measurements on impure materials. A third was to find methods to measure conductivity and transport in powders or in small crystals. 1. Structure Selection The basis of the structural approach^3-' is to concentrate the search on Type II and Type III ionic conductors since it is unlikely that Type I conductors can sustain the level of conductivity required for high energy density batteries. Type I conductors are solids in which the concentration of defects is small, generally less than 1019 cm"3; Type II are highly nonstoichiometric compounds which usually have a broad solid solution composition range; and Type III are compounds which have structures with many vacant sites of equal or nearly equal energy which mobile cations can occupy. This approach was well validated in the first year1 s program since many of the solids which satisfied these structural requirements gave evidence for some ion mobility at ambient temperature. As a result of the first year1 s work it was possible this year to make the structural criteria more restrictive. Nonstoichiometric compounds frequently contain ions in more than one valence state. A typical example are the lithium vanadates which were studied in the first year1 s program. Multiple valence states resulted in considerable electronic conduction which made it difficult to assess the amount of charge transported by ions. In the materials prepared for this study the electronic conduction was reduced by incorporating charge compensating ions without destroying the desired structure. Last year, several open framework, low density compounds were studied which were relatively stoichiometric and contained only a small concentration of defects. Such solids were tested with the hope that they would exhibit some motional narrowing of the NMR spectrum, thus indicating ionic motion, and that the level of mobility could later be enhanced by substituting foreign ions which would introduce the necessary defects. Examples for such compounds are NaBO2 and Na2BeF4. However, since none of these compounds gave evidence for potential ion motion, screening this time was limited to compounds which were known to contain defects. In selecting structures for a more detailed study, use was made of three sources. (1) The list of structures for which a preliminary study had already been made in the first year of the contract. Here the objective was to pick the most promising candidates of the previous year -- those which had successfully passed preliminary NMR screening -- and to synthesize them in a purer and/or nonstoichiometric form. (2) A large list of promising structures that had been compiled some years ago. From this list a small number had been selected for study during the first year of the contract. Some structures from the larger original list were reconsidered during this contract period, even though they had been passed over during the first year. (3) New promising structures obtained from a search of the recent literature. In selecting a final group of candidates from these lists, the following crystal-chemical criteria have been applied: (1) The candidate structures must have either (a) one-dimensional tunnels, (b) two-dimensional sheets of interconnected tunnels, or (c) a three- dimensional framework of interconnected channels. Such channels should be wide enough to allow for the motion of monovalent ions such as Li1 , Na1+, or K1+. (2) If a one-, two- or three-dimensional path of ionic motion is available in a particular structure, a search was made for other clues which might indicate " looseness" of the monovalent ion, or empty sites in the structure toward which an ion could migrate. Such clues include: (a) Abnormally large cation-anion distances for the migrating cation. (b) Abnormally large x-ray temperature factors, meaning larger than normal thermal vibrations for the diffusing cations. (c) Availability of vacant sites or structural holes toward which the cation could move. (d) Potential of creating nonstoichiometric solid solutions or vacant sites in compounds where the path of easy ionic motion is "plugged up" with interfering cations. (3) The physical and chemical properties of the candidate compounds should be of such a nature that practical applications are possible. This requirement eliminates from consideration large groups of compounds: (a) The solid electrolyte should be chemically resistant toward attack by liquid alkali metals. Many transition metal ions would be reduced under such conditions and are therefore unsuitable. (b) The candidate compounds should be electronic insulators. This requirement eliminates from consideration many transition metal ions with partly filled d-shells, e.g., tungsten or vanadium bronzes. (c) Potential solid electrolytes should be chemically stable under ambient conditions; for example, they should not be too hygroscopic. This requirement makes poor candidates out of many lithium- or sodium-rich oxide phases which might otherwise be attractive candidates. (d) Potential candidates should be capable of being fabricated into dense compacts. The structures and compounds that were finally selected for experimental investigation were determined by the following procedure. A list of candidates with supporting documentation was generated from a survey of the literature and then submitted to NASA-Lewis Research Center for selection of specific compositions to synthesize and evaluate. Some candidates were unacceptable because there were reasons to believe difficulties would be encountered in their utilization in practical battery systems. One reason was ease of fabrication. Although Li3AlN2, Li3BN2, and salt-impregnated zeolites had given strong evidence for large ionic motion at ambient temperature, they were rejected because they are hygroscopic and would be difficult to fabricate. Compounds which contained iron, cobalt, vanadium, and similar ions were excluded because they probably would be unstable toward oxidation- reduction reactions. The fluorite arrangements of NaPrF4 and NaYbF4 are probably good fluoride ion conductors but they were not investigated further because cation conductors are specifically desired. The encouraging result obtained for lithium tantalate was not followed up because a study of the tantalate system was to be undertaken at the National Bureau of Standards. 2. Synthesis and Fabrication of Materials One of the objectives of the preliminary search was to screen a large number of compounds which were to be selected from as many different structure types as possible. For this reason it was not possible to devote much time to refining the synthesis, and many of the measurements were made on impure multiphase materials. The results showed that impure powders could be screened for ion motion by NMR but that pure single-phase material was essential for the evaluation of conductivity and transport. Particular emphasis was placed on attempts to make nonstoichiometric defect solid solutions. Many structures have potentially conducting channels or planes which are filled to capacity with alkali metal ions. If vacancies could be introduced into such tunnels or planes, it could provide a passage for the migration of alkali metal ions. The synthesis of the various compounds was normally done by established methods suggested in the literature. Frequently these methods had to be modified to obtain higher phase purity or to make nonstoichiometric solid solutions. The phase purity of the various compounds was checked primarily by x-ray diffraction, supplemented by microscopic observation, especially for the glassy phases. An effort was made to identify all phases, including the impurities. 3. NMR Screening The screening of solids for ionic motion by observing motional narrowing of nuclear magnetic resonance spectra was accomplished with the same tech- niques described previously. (2) The measurements were made by I. Chung and H. S. Story in the Physics Laboratories of the State University of New York in Albany. Compounds in which the charge carrier was predicted to be a lithium or sodium ion were screened by measuring the.NMR spectra at two or more temperatures. Line narrowing with increasing temperature was evidence for ionic motion and was a necessary but not sufficient criterion for ionic transport. The spectra"were measured with a Varian Associates wide-line NMR spectrometer with "signal averaging. 4. Conductivity and Transport Measurements The significant properties for the evaluation of solid electrolytes are ionic conductivity and transport number. These measurements usually require either large single crystals or polycrystalline compacts which are dense and single phase. Since the growth of large crystals and the develop- ment of fabrication procedures for densifying new materials are difficult and time-consuming processes, the search for new electrolytes would be expedited by procedures which can be applied either to powders or to small crystals. A number of methods which were tested for achieving this objective are described below. (a) Two-phase measurement of conductivity Professor Charles W. Tobias suggested it might be possible to determine the conductivity of a powder by suspending it in a solution which contained the same conducting ion from measurements on the total conductivity of the two- phase system. Two experiments were performed by S. P. Mitoff and R. N. King to test the feasibility of this idea. The experiments showed that the two-phase method could riot be used in its present form to obtain reliable conductivities of suspended powders. The tests were carried out by measuring the conductivity of a solution of Nal in propylene carbonate containing a suspension of 60 mesh Monofrax beta- alumina powder. The measurements were compared with similar measure- ments on a suspension of Pyrex glass with the same particle size. The same results were obtained from two experiments. The second experiment was made with particular care to insure the solids were wet by the liquid by heating the suspension to 95°C under a vacuum of 30 inches of Hg. Both experiments showed the Pyrex glass and beta-alumina increased the resistance of the liquid electrolyte by the.same amount (e.g., from 1320 ohms to 3350 ohms, four-probe AC measurements). The measurements showed that both the Pyrex glass and beta-alumina behaved as insulators and the increase in resistance was due to the volume of the suspended solid. It was concluded that the conducting properties of the two-phase system are determined by the solid-liquid interface and cannot easily be used to obtain information about the suspended solid. No further work was done on this method as it seemed unlikely that it could be used in its present form to measure the conductivity of an unknown powder. (b) Microwave measurement of conductivity of powders The bulk conductivity of powders can be measured at microwave frequencies because at very high frequency the individual crystallites are capacitively coupled and the results are determined by the intrinsic properties of the solid. Microwave techniques are frequently used to measure semiconductor and dielectric materials, but to our knowledge they have not been used for the study of ionic conductivity in loose powders. The validity of the method was established by measuring the conductivity of beta-alumina. In addition a few preliminary measurements were made on two candidates synthesized in this study (W-doped LiNb6O15F and Li2Ge7O15). At the request of NASA- Lewis Research Center, the microwave screening studies were discontinued and the candidate materials were sent to their laboratories for dielectric loss measurements, '^b) since both were high-frequency measurements, this was done to avoid duplication of effort and allow the General Electric study to concentrate on the measurement of micro-specimens. A brief description of the microwave equipment and validation measurements is given below. The conductivity and dielectric constant of powders was measured in the range of centimeter waves by the " hollow-pipe" method introduced by S. Roberts and A. von Hippel(2c) [j. Appl. Phys., IT, 610-616 (1946)] for dielectric measurements. The specimen is contained in a cavity attached to a coaxial line. A transmitter radiates waves of a given frequency which are reflected by a metallic short which terminates the cavity. Standing waves are set up and measured by a probe traveling along a slot in the pipe parallel to its axis. The conductivity and dielectric constant are calculated from the standing wave ratio (SWR) in air and the distance of the first node from the surface of the specimen. The conductivity and permittivity are related by a transcendental equation which cannot be solved explicitly in terms of the measured parameters. The desired quantities were obtained by using a pro- gram written by G. H. Glover of our Laboratory in which a computer inverts the transcendental equations by successive interpolation. The specimen chambers were made of brass and the measurements were made over the frequency range 2 to 4 GHz. The microwave method was tested by measurements on loosely packed powders (approximately < 100 mesh) of Monofrax H beta-alumina. The resistivity and dielectric constant of beta-alumina are given in Tables I and II. The results show powder conductivities were obtained which are close to accepted values measured on single crystals and dense ceramic specimens. The larger resistivity of the specimen that had not been dried agrees with the well known effects of water vapor absorbed in beta alumina. The frequency dependence indicates some dispersion but the method appears satisfactory for screening purposes. More work would be required to establish the uniqueness of the param- eters deduced by the inversion procedures. The hyperbolic functions are multivalued and measurements with a single thickness of specimen may give ambiguous results. The problem can be solved by measurements on specimens of different thickness. If the conductivity is calculated from two different thicknesses of material, only one pair of values will satisfy both sets of experimental results. TABLE I Microwave Conductivity and Dielectric Constant of Beta-Alumina Powder at Room Temperature sr 3.00 2.59 2.94 3.31 3.83 4.03 5.31 6.43 7.83 Frequency (GHz) 4.00 3.75 3.50 3.25 3.00 2.75 2.50 2.25 2.00 Not dried p (ohm -cm) 756 1620 1390 2420 2088 7917 5694 5744 — er 12. 1 8.8 10.4 11. 3 12.6 2. 73 3.72 4.25 — Dried p (ohm -cm) 810 I 1388 J 1732 2 2052 i 2027 I 1942 1605 i 1424 1 1170 8 TABLE II Temperature Dependence of Microwave Conductivity of Dry Beta-Alumina Powder Temp (°C) p (ohm-cm) er 24 810 3.007 52 614 3.012 60 560 2.960 70 497 2.976 80 473 2.828 90 438 2.810 100 413 2.830 (c) Conductivity, transport, and pulse measurements on micro-specimens The following methods were developed to determine the ionic transport in small specimens. They can be used on glasses or poly crystalline ceramics, coarse grained powders, and single crystals. The techniques are still being improved but the methods have already given useful information that could not have been obtained with conventional methods. Key contributions to the success of these methods were made by Dr. G. C. Farrington for the techniques of measuring conductivity in nonaqueous solvents and with fast pulses, and by Mrs. Ethel Fontanella for developing methods for potting solid electrolytes and assembling conductivity cells. The first apparatus was a Teflon polymer cell in which specimens 2 to 4 mm in diameter and about 1 mm thick were measured successfully. The method next was applied to specimens with dimensions between 0. 5 and 1 mm by potting the specimen in an epoxy cell (microcell). Some measurements were made with conven- tional four-probe DC and AC bridge methods, and particular success was obtained with fast pulse measurements. i. Teflon polymer conductivity cell. The first conductivity cell is shown in Fig. 1. Above is an overall view of the assembled cell and below a disassembled cell. The cell is made of two Teflon polymer blocks, 1 1/4 x 1 1/4 x 1 inch and 1 1/4 x 1 1/4 x 7/8 inch, which are drilled with cavities and clamped together to form compartments into which platinum or silver electrodes are inserted. The compartments are separated by an epoxy disk which contains the solid to be measured. The epoxy disk is sealed into the cell with a pair of Viton or silicone rubber " O" -rings. Each electrolyte compartment contains about 2 ml electrolyte solution and a variety of solvents and electrode systems were used to study different ions and for potentiometric measurements. Fig. 1 Conductivity cell. Overall and disassembled views are shown. The specimen is a cross section of a beta-alumina tube tested to validate the pulse method for conductivity measurement. 10 The sample to be tested is potted in a specially formulated epoxy resin in the form of a disk, 16 mm diameter and usually about 0. 5 mm thick. The disk is cast in a silicone rubber mold and curing techniques have been developed which produce disks without gas pockets. The final step in specimen preparation is to grind the disk until the solid electrolyte is exposed on both sides. The potted specimens are helium leak tested before assembling in the cell. Two methods are used to avoid leaks between the specimen and the epoxy potting matrix. Such leaks frequently develop during polishing to expose the surfaces of the solid electrolyte. The solid electrolyte surface may be protected by clamping the specimen between silicone rubber disks before submerging in epoxy. Alternatively, opposed small areas of the specimen are coated with RTV-112, a pourable silicone rubber sealant, cured for 24 hours, then imbedded in epoxy contained in a silicone mold. The cured RTV can be stripped off the surface after the epoxy resin has hardened. The latter procedure can be done under the microscope and is particularly useful for small samples. The disk is clamped between the " O" -rings in the cell and the complete assembly tested for helium leaks. The cell then is placed in a dry box, the electrolyte solutions are added, and the electrodes are inserted. Electrolyes of lithium perchlorate or sodium trifluoromethane sulfonate dissolved in propylene carbonate and acetyl nitrite are stable and do not react with the components of the cell. The cell was tested by preparing potted disks of ceramic beta-alumina and single-crystal beta-alumina. Disks were hand lapped to about 0.5 mm and to less than 0. 2 mm with a jewelers lathe. This thickness sets the lower limit of crystal size that can be tested by this method. ii. Microcell. The assembly shown in Fig. 1 was not satisfactory when specimens smaller than 0. 5 mm were measured with the pulse technique that will be described later. This was due to the fact the epoxy matrix has a small but finite conductivity and capacitive and polarization effects at the liquid electrolyte-solid interface dominated the response of the samples with small dimensions. The problem was solved by constructing microcells of the type shown in Fig. 2 in which the liquid electrolyte-solid interface is reduced to the sample size. In essence, a cell is built around each sample by casting epoxy bubbles on both sides of the potted specimen, then tunnels are drilled to contain the liquid electrolyte and platinum leads. The liquid electrolyte is introduced into the cell with a hypodermic needle. iii. Pulse technique for conductivity measurement. The procedure developed to measure ionic conductivity is to observe the structure of a square wave constant pulse in order to separate the IR drop in the solid electrolyte from interfacial polarization. 11 Fig. 2 Microcell for small specimens. Also shown is a potted sample before cell assembly. The liquid electrolyte is introduced with a hypodermic needle. 12 The experimental apparatus consists of a 2N3904 transistor acting as a fast response switch in series with a ca 20 Kohm resistance and a 22 1/2 volt battery. Switching is effected by driving the transistor base with a 0 to +4v square wave pulse train supplied by a PAR Model 175 universal programmer. The programmer also triggers a Model 7B52 delayed sweep time base in a Tektronic Model 7704A oscilloscope. Vertical oscilloscope input is provided by a dual trace Model 7A12 amplifier which permits simultaneous monitoring of the potentials of the two sensing electrodes in the Teflon polymer electrolysis cell. The interface polarization was displayed by observing the rise and decay of a 3 x 10~4 amp current pulse introduced as a 0. 05 sec square wave. The voltage drop due to polarization was manifested by approximately exponential slow components superimposed on the leading and trailing edges of the square wave. The resistivity of the beta-alumina calculated from the initial rapid decay was found to be 1550 ohm-cm. With this apparatus it is possible to apply constant current pulses of varying amplitudes to the two working electrodes in the Teflon polymer electrolysis cells while simultaneously monitoring either the total pulse shape or magnified pulse structure at initiation or termination. Pulses of 200 msec to 1 msec duration have generally been used. These have had rise and fall times into a 2000 ohm dummy resistance of less than 0. 4 (jsec. In an actual measurement, a series of current pulses is applied to the electrolysis cell which is connected in series with a precise 2000 ohm General Radio resistor. Voltage drops across the sensing electrodes and the standard resistance are recorded. Typical currents are in the range of 15 to 50 namp. As has been described previously, the pulse measurement technique relies on the concept that the portion of the total applied cell voltage which arises from electrolyte iR drop decays almost instantaneously upon cessation of current flow while that portion of cell voltage which arises from interfacial polarization decays at a much slower rate, typically over a period of 10 to -50 msec. If a sharp enough current pulse can be applied to a cell and sensitive enough equipment is used to carefully examine the resulting shape of the total applied voltage, absolute magnitudes of iR drop and interfacial polarization can be determined. A four-electrode technique was used in these experi- ments to separate any electrochemical polarizations at the electrodes where current is applied from the polarizations at the solid electrolyte/liquid electrolyte interface. 13 III. VALIDATING EXPERIMENTS A. Bridge Measurements The reliability of the conductivity techniques was tested by measurements on beta-alumina ceramic specimens of known conductivity. The room temper- ature resistivity of freshly prepared ceramic beta-alumina, measured by the four probe technique, was 2490 ohm-cm at 10 Hz and 66 ohm-cm at 300 kHz. Experience has shown these values may increase slightly (25$) due to absorption of water from the atmosphere. The beta-alumina was cut and ground to a regular parallelepiped with area equal to 0. 8 x 0. 6 = 0. 48 cm2 and thickness of 0. 075 cm and potted in epoxy resin. Measurements were made in a Teflon polymer cell with coiled Pt electrodes and 1M NaClO4 in propylene carbonate. Initial values were obtained at 1000 Hz with an AC bridge. The resistance of the cell with no separator was 496 ohms and that of the blank epoxy disk without beta-alumina about 106 ohms. The apparent resistance of the beta-alumina membrane, after subtracting that of the liquid electrolyte, was then found to be 2700-496 = 2204 ohms, corresponding to a resistivity of 2204 x °'48 - 15, 000 ohm-cm.0. 075 The large resistivity calculated frbm the cell measurement, approximately seven-fold greater than the four-probe DC value, is evidence of significant polarization at the solid-liquid electrolyte interface. B. Applications of Pulse Technique to Beta Alumina The pulse technique was validated by measuring the ionic conductivity in a well-characterized sample of beta-alumina. The beta-alumina ceramic was cut and polished to expose a disk 2 mm in diameter and 0. 474 mm thick. Figure 3 shows the overall structure of the cell voltage response to a constant current pulse through beta-alumina separating two-cell compartments filled with 0. 1M NaClO4 in propylene carbonate. The numbers in the upper- and lower-left corners are the vertical scale factors in mV/cm for the upper and lower traces, respectively. Sweep time in Usec/cm is recorded in the upper or lower right-hand corners. Total cell voltage between the two sensing electrodes is the difference between both traces. Note in Fig. 3 the sharp rise upon onset of pulse and sharp fall upon termination. The termination region is magnified in the lower photograph. A sharp drop occurring within a fraction of a microsecond occurs in the voltage of each electrode. Only the electrode response recorded in the upper trace includes the effects of the beta-alumina; the lower trace arises from the sensing electrode separated from ground solely by propylene carbonate 14 Fig. 3 Cell voltage response to constant current pulse through beta-alumina. Lower trace is expansion of termination region. 15 solution. The sharp voltage drops are interpreted as measurements of iR loss whereas the slower potential decay of the upper trace reflects decay of interfacial polarization at the beta-alumina surface. An ionic conductivity of 1100 ohm-cm is calculated from these data after correction for the resistance of the propylene carbonate solution. Since the conductivity and area of the candidate conductors frequently are much smaller than that measured above, experiments were undertaken to validate the method with a smaller specimen of beta-alumina. An attempt was made to determine the resistance of a sample of poly- crystalline beta-alumina having an area exposed of 0. 0020 cm2, or about 1$ of that of any sample previously measured. The cell response was the same as that exhibited by a pure epoxy disk under similar conditions. The sensitivity of the pulse technique for small specimens was increased by maximizing the ratio of sample area to disk area. With a minute sample of beta-alumina pulse measurements with the microcell described in the preceding section gave p ^ l O O O ohm cm, in agreement with accepted values. An additional complication stems from the fact that the epoxy disk between two conducting liquids acts as a capacitor. When the conductivity between the cells is high, this has little effect on these measurements. However, when the conductivity is low, the entire cell responds as a capacitor. The constant current source then ceases to function properly and the pulse technique becomes useless. The microcell, in which the nonconductive area is minimized, solved this problem as well. The possibility was explored of using a potentiometric method to establish ionic conductivity in specimens which are extremely small or resistive. The liquid electrolyte was replaced by solutions which were 0. 2M AgClO4 + 0. 1M NaClO4 and 0. 002M AgClO4 + 0. 1M NaClO4 on opposite sides of the beta alumina sample. The potential developed between Ag electrodes was 0. 131V, in reasonable agreement with the theoretical value Unfortunately, the same result was obtained when the blank epoxy disk was substituted for that containing beta-alumina, and we conclude the epoxy resin or Viton rubber are too conducting for this to be a practicable method. 16 IV. RESULTS AND DISCUSSION OF RESULTS A. LiNbeOisF-Type Compounds 1. Structure and transport model The tunnel structure found for LiNb6O15F(D is illustrated in Fig. 4. The building blocks are made up of pentagonal NbX7 bipyramids, sharing edges with five NbX6 octahedra. These building blocks are joined together at the corners in three dimensions, leaving hollow tunnels in which the Li1+ ions are presumed to occupy two possible sites. * Either site would yield six to eight Li-O Fig. 4 The structure of LiNb6O15F. (Li1 ions are not shown. ) Taken from Ref. 1. *Due to their low scattering power, the Li atoms cannot be located easily by x-ray methods. The Li-sites have been inferred from the knowledge of the rest of the structure. 17 distances between 2.5A and 3. OA. These distances are considerably larger than conventional Li-O bonds which are of the order 1. 95A to 2. 30A. It would appear, therefore, that these tunnels are wide enough for migration of Li1 ions along the tunnels or b-direction of Fig. 1. In preliminary NMR screening(2) motional narrowing at -26°C was observed for 7Li in LiNb6O15F, indicating at least local ionic motion of the Li1 + ions. If the Li ions occupied fully one of the two Li sites, with the other type of channels completely empty, ionic conductivity would not be expected in stoichiometric LiNb6Oi5F. However, one could image nonstoichiometric variants (listed below) with either lower or higher Li content, made by substituting ions such as Ti4+, Mo6+, or W6* and adjusting the O/F ratio. It is quite possible that both Li sites are partly occupied by Li1 ions. In this case high ionic mobility might be expected even for stoichiometric LiNbeO^F. 2. Synthesis (a) LiNb6Oi5F and Lio.8Nb5. 6W0. 4Oi5. 2F0.8 LiNb6O15F and Lio.8Nb5.6Wo.4O15.2Fo.8 can be prepared by heating the ap- propriate mixtures of LiF, M^Os and WOs in sealed platinum tubing (as a loose powder) or a sealed platinum crucible (in the form of compressed pellets). For one set of samples the heat treatment was 2 days at 1200°C in a muffle furnace. A second set of samples was heated for 1 day at 950°C. The resulting products for both temperatures and compositions were single phase and of the LiNb6O15F type. No impurity peaks could be identified by x-ray dif- fraction. The crystals tend to be blade -like and up to 2 mm in length. (b) Lip. 9Nb5. 8Mo0. 2Oi5. i^o. 9 and Li0. 86Nb5. 72Mo0. 2sOi5. uFp. 86 Mo-substituted phases of the above compositions could be prepared in the same manner as described above for LiNb6Oi5F, except that here the samples are heat treated at lower temperatures: 1 day at 950°C. X-ray diffraction data showed that these phases were of the LiNbeOi5F type. No impurity peaks could be detected. (c) Attempted Synthesis of Li0. 8Nb5. 6Mo0. 4O15. 2F0. 8 and Lip. 7Nb5. 4Mo0. Attempts were also made to synthesize LiNb6O15F-type compounds of compositions Li0. 8Nb5. 6Mo0. 4Q.5. 2F0. s and Li0. 7Nb5. 4Mo0. e^is. 3Fo. ?• How- ever, these two compositions gave a two-phase mixture: (a) a coarsely crystalline LiNb6O15F-type phase with blade -like crystals up to 2 mm, (b) a fine needlelike phase of a darker color. 18 (d) Ti -Substituted Phases Attempts were made to prepare a LiNb6Oi5F-type phase of composition Li1>2Nb5.6Ti0. 4O14>8F1>2- The synthesis proceeded as described for LiNb6Oi5F. A compound of the desired LiNb6O15F structure type could not be obtained, and it is tentatively concluded that Ti4+ cannot be substituted into the LiNb6O15F structure in significant amounts. 3. Evaluation Strong motional narrowing of 7Li in stoichiometric LiNb6Oi5F was previously reported. (2) The dependence of line width on temperature indicated the lithium motion was probably localized or short range. NMR spectra of LiNb6O15F doped with molybdenum and tungsten are shown in Figs. 5 (a), (b), (c), and (d). The 7Li spectrum, measured at room temperature and 186°C, shows strong motional narrowing. The line widths of the doped compounds are more narrow than in stoichiometric LiNbgOisF (Table III), showing that substitution of Nb by Mo and W increases the lithium mobility, probably because more vacancies and disorder were introduced in the tunnels. No motional narrowing of the fluorine NMR spectrum was detected over the same temperature range. The absence of a Knight shift in the stoichiometric compound was previously reported to show the absence of electronic conductivity. It is believed the same is true for the doped compounds because the composition of the phases was adjusted to maintain the niobium in the pentavalent state. TABLE III NMR Spectra of LiNb6Oi5F Compounds Sample C-5a Compound Li0. 9Nb5. 8Mo0. . iF0. 9 C-9a . 9 LiNb6Oi5F Nucleus Temp (°C) rLi rLi 19F 22 186 22 186 22 186 22 186 22 183 Width, Gauss 1.8 ± 0 . 5 0. 17 ± 1.0 4.0 ±1.0 4.0 ±1.0 2.0 ± 0 . 4 0. 18 ±0.04 4.0 ±1.0 Not measured 3.0 ± 0 . 2 0. 6 ± 0. 1 19 I ' I I 11 • I ' ' M I I 11 I '-' 11' ' • ' ' I.I.1.1 I 11 I.I > I Fig. 5 (a) NMR spectrum of 7Li in Lio. gNbs. 8Mo0. 2QL5.iFo. 9 at room tempera- ture. The peak width is 1. 8 gauss. LiNb,0._F: Ko . I . . . . I . . . . I . . . I . . . . I I . . . . I . . . . I . Fig. 5(b) Above at 186°C. The peak width is motionally narrowed to 0. 17 gauss. 20 T' ' 1 ' ' ' ' I Fig. 5(c) NMR spectrum of 7Li in Lio.8Nb5. 6W0.4Oi5. 2F0. s at room tempera- ture. The peak width is 2. 0 gauss. Fig. 5(d) Above at 186°C. The peak width is reduced to 0. 18 gauss. 21 The crystals that were available were too small for reliable transport measurements at the present stage of development of our microcell method. Attempts were made to measure the conductivity in molybdenum -doped single crystals (Li1-xNb6_2xMo2XOi5+xFi-x with x < 0. 2). In order to prevent leaks" about the edges, it was necessary to pot the crystal completely and expose a circular area only 0. 25 mm in diameter for measurement. The growth habit is not known so individual crystals were potted with their major face parallel and perpendicular to the applied potential. The first set of measure- ments, which were made in the Teflon polymer cell, gave responses that were the same as expected for the epoxy disk. The measurements were repeated with similar results in a microcell to reduce interface effects (Fig. 6). It is estimated that the conductivity is probably less than 10~5 ohm"1 cm"1 at room temperature. Dielectric loss measurements were made at NASA-Lewis Research Center on three of the lithium niobate compositions (Table IV). Neither the stoichiometric compound (C-l) nor the tungsten-doped material (C-3) gave indication of resonance peaks over the temperature range of 100° to 500°K. Titanium -doped lithium niobate (C-4) exhibited resonance peaks which were taken as strong evidence that the relaxation responsible for the resonances is that of mobile ions. The AC conductivity measured at 106 Hz was as follows: TABLE IV AC Conductivity (ohm"1 cm"1) Sample _ Composition Room Temp 500°K C-l LiNbeAsF 2 x 10~7 2 x 10"6 C-3 Li0. 8Nb5. 6W0. 4Qi5. 2F0. a 5 x 10-7 6 X 10"6 C-4 Lii.gNbs.eTio^OH.sFi. 2 8 X 10-6 6 X 10-4 In summary, lithium niobium oxyfluoride appears to be a moderate to poor conductor of lithium ions at low temperature, and the conductivity is improved by doping. B. Hollandite-Type Compounds 1. Structure and Transport Model Hollandite is a tunnel structure which takes its name from the mineral hollandite, approximately BaFeMn7O!6 . K1+, Na1+ and Pb2+ can substitute for Ba2 and Al for manganese or iron. The structure was first described by Bystrom and Bystrbm(3) and is illustrated in Fig. 7. In the structure, MnO6 octahedra and FeO6 octahedra randomly form an open framework which leaves empty channels. Approximately half of the available sites in the channels are occupied by Ba2+ ions. It has been postulated^3) that these Ba2+ ions are strung out along these tunnels in a regular manner, such that each 22 S3 Fig. 6 Response of Mo-substituted LiNbeO^F cry stalin microcell. The lower trace is an expansion of the termination region. 23 Fig. 7 The structure of hollandite projected on the (001) plane. After Ref. 3. Ba2 ion is followed by a vacant site and each vacant site by a Ba2+ ion. These Ba2 -D-Ba2+- .... chains are randomly arranged in the structure, so that there is only a one-dimensional order in the tunnels. It has also been postulated that the vacant sites between Ba2+ ions can be partly occupied by water molecules which are always present in the mineral. In recent years a large number of new hollandite-type compounds have been reported'^, 5) in which alkali metal ions (especially K1 and Rb1+) com- pletely replace the Ba2+ ions inside the tunnels. Such K1+ or Rb1 -hollandites can be completely stoichiometric so that all sites within the channels are occupied, and no vacancies exist, e. g., K2ZnTi7O16, K2Ga2Ti6O16, Rb2MgTi7O16, Rb2Al2Ge6O16, etc. For some systems, both stoichiometric and non- stoichiometric phases can be prepared. For example, in the system, stoichiometric K2Al2Ti6O16 and nonstoichiometric K^e GIB can be prepared, both with the hollandite structure. (4) Stoichiometric hollandites have also been made with Ag1 , T11+ or Na1+ in the tunnels; '^) however, Cs1+-hollandites can only be made non- stoichiometric. (4) Apparently the large size of the Cs1+ and Ba2+ ions does 24 not permit complete filling of the hollandite channels. For Ba2 -hollandites, the repulsion between two adjacent Ba2 ions is probably another reason why attempts to synthesize stoichiometric Ba-hollandites (such as Ba2Mg2Ti € O16) have so far been unsuccessful. 2. Synthesis The hollandite samples were of composition Ki.eMgo^Tiv^Oie and were obtained from the NASA-Lewis Research Center. 3. Evaluation o The hollandite boule was described as poly crystalline with preferred orientation of crystallites with tetragonal c-axes parallel to the boule axis. The resistivity at room temperature measured by dielectric loss on similar material was 80 ohm-cm. On the other hand, the room-temperature resistivity measured with four-probe DC was 104 to 105 ohm-cm. No attempt was made to determine motional narrowing of the potassium NMR spectrum because the nuclear moment is very small and the potassium peak probably would not have been seen above background. Four hollandite specimens were cut from the boule for conductivity and transport measurements. Each had a cross-sectional area of about 0. 12 cm2 and was 0. 10 cm thick. Two were cut parallel and two were cut perpendicular to the boule axis. Identical results were obtained with the two pairs of duplicate specimens. c The conductivity of hollandite was measured in the Teflon polymer cell. The top trace of Fig. 8 shows the voltage response to an 18 (jamp, 200 msec pulse through a solution of 0. 05M tetrabutyl ammonium tetrafluoborate, 0. 02M KC1C»4 in propylene carbonate, without an epoxy disk or sample present. The center trace shows the response to the same current pulse with the cell containing the sample of hollandite cut with the conducting tunnels (c-axis) parallel to the pulse direction. Note the well-defined iR drop upon pulse termination. We have calculated a hollandite ionic conductivity of 200 ohm-cm from these plots after correction was made for the resistance of the propylene carbonate solution. In contrast, the bottom trace of Fig. 8 shows the cell response when the same current pulse was applied to a hollandite sample in a direction perpen- dicular to the conducting tunnels. The cell impedance is so high that less than 1 (j.amp current flows. No sharp iR drop is seen in the magnified trace because the total cell impedance is so high that the cell responds like a capacitor shunted by a large resistance. Sample resistance perpendicular to the tunnels is estimated to be greater than 105 ohm-cm from these traces. 25 Fig. 8 Hollandite. The upper trace is voltage response with no specimen present: 18 namp - 200 U sec pulse through 0. 02 M KC1O4. Middle trace is response to hollandite oriented with tunnels parallel to field. Lower trace is response when tunnels are perpen- dicular to the field. 26 The pulse measurements having shown that there is large ionic con- ductivity in a direction parallel to the tunnels in hollandite, an attempt next was made to determine the potassium transport down the tunnels. The hollandite slab was placed between two propylene carbonate solutions, both 0. 1 Min tetrabutyl ammonium tetrafluoroborate one saturated with KC1O4. The propylene carbonate was dried over activated, alumina. The cell was filled in a dry box, then closed tightly to prevent contamination by water before removing for measurement. The hollandite specimen was polarized between two platinum electrodes maintained at 5V potential difference. The cell developed an initial current greater than 1 |aamp which decayed to 0. 09 (jarnp in 5 minutes after trans- porting a total of 2. 7 x 10~5 coulombs. After 20 minutes the current was 0. 05 namp and 7. 4 x 10~5 coulombs had been passed. The current continued to decay with time and the experiment was discontinued after one hour 52 minutes when i = 0. 03 |j.amp and Q = 2. 3 x 10~4 coulombs. Pulse measure- ments were made in both directions throughout the experiment and we con- tinued to observe the same responses as previously reported. No attempt was made to measure the transport number by analyzing for potassium since the maximum potassium that could have been transferred was only 2 x 10~9 eq. There are about 1015 tunnels per cm2 of the c-face of hollandite. The large current that was observed at the beginning of the experiment corresponded to a transport of about 2 x 10 ~14 coulombs/cm2, approximately equal to that calculated if a few potassium ions entered or left each tunnel. The current then decayed to the relatively small value of 0. 2 |aamp/cm2 which may represent migration down the tunnel. We conclude there is high short-range conductivity in potassium hollandite and this property is measured by both the DC pulse method and by dielectric loss. However, the transport and DC measurements show that long-range conductivity at room temperature is small and probably is determined by the slow migration of potassium ions down the tunnels. C. Cordierite-Type Compounds 1. Structure and Transport Model The structure of the mineral cordierite (essentially MggA^SisOj^. xH2O) is another tunnel structure which is of interest as a potential solid electrolyte. The structure was first studied by Bystrom(6) and later by Gibbs, ^ Smith and Schreyer. (8) it is made up of a three-dimensional tetrahedral network of the composition Al4Si5O18. The Mg2+ ions occupy octahedral sites, and varying amounts of water are located in the wide channels running parallel to the c-axis. The structure is illustrated in Figs. 9 and 10 giving a good view of the channels which are lined with oxygen ions and have a radius of 27 75^9 Fig. 9 The structure of cordierite, viewed down the c-axis. After Ref. 6. \ c T Fig. 10 Cross-sectional view of a tunnel in the cordierite structure. After Ref. 8. Note that the tunnels are large enough to accommodate not only water molecules but even the larger argon atoms with a radius of 1. 92A. 28 1. 4A at the narrowest neck and 2. 2A at the widest. The channels can be completely free from any water, and in fact, the compound Mg^^SisOis exists as a stable phase in the ternary system MgO-AlgOa-SiQj. See W. Schreyer and J. F. Schairer. ^' It is known from the analyses of cordierite minerals and from recent crystal chemical studies that alkali ions such as Na1 can occupy to some extent the wide channels. Na1 ions are expected to be too small to occupy the center of the tunnels and would probably be located at the edge of the channels. Larger ions would probably occupy a more central position in the tunnels. It is possible ionic conductivity may result if alkali ions are introduced into these channels. The situation is expected to be especially favorable for K1 and Rb1+ ions. Recently cordierites of the compositions NaxMg2 (Al4_xBexSi5O18) have been synthesized where x can be up to 0. 55. Here the substitution Al3 = Be2++Na1 was used to introduce substantial amounts of Na1+ into the empty channels. See Provondra and Langer. '*") 2. Synthesis To synthesize alkali -rich cordierites, it was felt best to avoid systems involving the toxic element beryllium, such as the NaxMg2(Al4-xBexSi5O18) solid solution series. We have therefore tried to modify cordierite (MgA^SiKOis) by making other substitutions, namely: Si4+=Al3++K1+, Si4+=Al3^+Na1+, Mg2+=Li1++K1 + and Mg2+=Li1++Na1+. In the last two cases, the Li1+ion substitutes for Mg2+ and the larger alkali ions go into the channels. The following cordierite -type compositions were chosen for synthesis: Nao.2Mg2Al4.2Si4.aOi8, KO. 2Mg2Al4. 2Si4<8O18, Nao.4Mg2Al4. 4Si4.6Qi8, K Rb0. 4Mg2Al4.4Si4>6Oi8, Nao.jjMg^aLio^A^SisOuj and The starting materials were NH4A1(SO4)2 • 12H2O, MgSO4 and the appropriate alkali sulfates. Ammonium stabilized Ludox colloidal silica suspension was used as a source of silica. The sulfates were first dissolved in hot water; then the appropriate amount of Ludox colloidal silica was added. The suspension was heated to dryness (at ~200°C) in a drying oven and then transferred to a platinum crucible and heated gradually from 300° to 1200°C in a furnace in air. The samples were kept at 1200°C for about 20 minutes and then withdrawn from the furnace. At this stage the samples are fine, white powders which x-ray diffraction indicated is made up of three phases: mullite (3A12O3 • SiO2), sapphirine (Mg7Al18Si3O4o) and cristobalite -related (or carnegieite-like) alkali silicate phases. At this stage no trace of a cordierite phase was detected. The powders were now heated at 1560°C for 20 minutes in air and cooled. In this way glasses were formed which were devitrified for two days at 1000°C. 29 The products were then x-rayed with the following results: Nominal Composition X-ray Results Nao.2Mg2Al4.2Si4.sOj8 100$ cordierite. No detectable impurities. Ko.2Mg2Al4.2Si4.sOi8 98$ cordierite. Possible trace (2$) of spinel. ~90$ cordierite. Some spinel and possibly p-spodumene. ~90$ cordierite. Some spinel and possibly p-spodumene. Nao.4Mg2Al4.4Si4.eOi8 ~-97$ cordierite. Possible trace of a-A^Os. Ko.4Mg2Al4.4Si4.eOi8 ~95$ cordierite. ~5$ a The excess a-A^Os may indicate that the real composition of the solid solution is somewhat below the nominal composition in sodium or potassium. Some sodium or potassium may be lost through volatilization during the firing of the samples, and a chemical analysis will be necessary to determine the real alkali content. Refractive index measurements of both the crystalline and glassy cordierite samples indicate that the alkali substituted cordierites have significantly lower indices than pure crystalline or glassy For the Rb analogue, significant amounts of a pure, homogeneous glass could not be obtained. In attempting to make such glasses, the reaction was so vigorous that a foamy, porous material was obtained instead of pure glass. The major components of the porous material were glass and crystalline spinel -type MgAl2O4. The same type of foaming also occurred in several runs with the Na and K-cordierites, but here such foaming could be avoided by the use of smaller samples. Our results for the Na1 + and K1+ substituted cordierites are encouraging, and indicate that limited substitution of the types: Si4+=Al3++K1+ or Si4+= Al3 +Na1+ can take place. The possible presence of p-spodumene (LiAlSigOs) in the Li1 -substituted samples, raises some doubt as to whether the substitution Mg2 = Li1++Na1 + really occurs. The identification of small amounts p-spodumene in these samples must be considered as tentative, since only the most intense (3-spodumene-type x-ray peak could be detected. 3. Evaluation Several attempts were made to observe the sodium NMR spectrum in sodium cordierite. The spectra were averaged for 30 to 40 hours and mea- sured at high r-f to increase the sensitivity. In no case was the sodium peak detected above the background. A typical NMR spectrum of Na-cordierite obtained by sweeping through the region expected for the sodium peak is shown in Fig. 11. The failure to detect any sodium could be due to many 30 1 I ' ' ' ' I ' ' ' ' I ' ' ' ' I ' ' ' I I ' I ' I I ' ' ' ' I ' ' I I I I ' I ' I ' ' ' ' I ' MMfw^v^^^ Fig. 11 NMR spectrum of sodium cordierite (Nao.4Mg2Al4.4Si4<6Oi8) at room temperature. The 23Na spectrum was accumulated for 40 hours. reasons: low sodium concentration, large quadrupole interactions, and/or the distribution of sodium ions in.two or more crystallographic sites. The response of potassium cordierite (K0-2Mg2Al4t2Si4>8O18) was tested in a microcell. The exposed specimen area was a disk with diameter 1. 5 mm thickness about 1. 5 mm. The shape of the complete pulse and the trailing edge at high resolution is shown in Fig. 12. The small rapid iR drop and the current flow indicates that potassium cordierite is probably a potassium ion conductor. The room-temperature conductivity is small, probably less than potassium hollandite. The results of the dielectric loss measurements on the cordierite samples are in Table V. None of the materials gave indication of resonance peaks over the temperature range of 100° to 500°K. D. NaNbO2F2-Type Compounds 1. Structure and Transport Model The structure of monoclinic NaNbO2F2 can be described as a layer structure made up of (NbO2F2)n~ sheets which are held together by Na1+ ions only. (There are no anion bridges connecting these sheets. )(H) The sheets are apparently made up of NbC>4F2 octahedra, linked to each other at the four oxygen corners. The structure is illustrated in Fig. 13. The eight .nearest Na-X distances in the irregular NaF6O2 polyhedron range from 31 Fig. 12 Voltage response of potassium cordierite in microcell. The trailing edge of the pulse is shown below. 32 TABLE V Sample and Composition B42as B49as B41as Na0.2 B48as Nao.4Mg2Al4.4Si4.eOi8 Room Temperature Conductivity Frequency 100$ cordierite 95$ cordierite; 5$ A12O3 100$ cordierite 97$ cordierite; 3$ A12O3 (Hz) 103 104 105 106 103 104 105 106 103 104 105 106 103 104 105 ohm"1 cm"1 4. 9 x 10~10 3 x 10-9 1.8 x 10-8 1. 1 x 10~7 1. 3 x 10-9 9 x 10-9 7 x 10-8 4 x 10'7 3 x ID"10 2 x ID'9 9 x 10-9 1.2 x 10-9 6 x 10-10 4 x 10"9 1.5 x 10-8 Fig. 13 The structure of NaNbO2F2 projected on the (010) plane. After Ref. 11. 33 2.221 to 3. ISA). The isotropic temperature coefficient for Na1+ (B = 2. 42) indicates larger than normal thermal vibrations. ' ' We believe that there is a chance of high Na1 ion mobility in this structure if vacancies can be introduced into the Na sites. For this purpose, the composition of NaNbO2F2 could be modified by changing the O/F ratio or introducing hexavalent ions into the octahedral site, e.g., Mo6 or W6^. One could similarly try to modify the composition of NaNbO2F2 by substituting Ca2+, Sr2+, Pb2+, Y3+ or Nd3+ ions for Na1+. 2. Synthesis There is probably little chance of Na1+ ion conduction in stoichiometric - With a view of making sodium -deficient NaNbO2F2 phases, the following reactions were tried: 0. 9NaF + NbO2F = Na0.9NbO2F1.9 0. 9NaF + 0. 9NbO2F + 0. !MoO3 = 0. 9NaF + 0. 9NbO2F + 0. 1WO3 - 0. 7NaF + 0. 1YF3 + NbO2F - Nao 0. 7NaF + 0. !NdF3 + NbO2F = Nao.7Ndo.iNbO2F2 0. 8NaF + 0. !CaF2 + NbO2F - Nao.8Ca0.iNbO2F2 0. 8NaF + 0. !SrF2 + NbO2F = 0. 8NaF + 0. !PbF2 + NbO2F = For most chemicals, reagent grade starting materials were used, but NbO2F had to be prepared according to the following procedure: Optical grade Nb2O5 (99. 95$) was dissolved in concentrated hydrofluoric acid in a platinum crucible. The resulting solution was filtered and the clear solution was evaporated to dryness. The powder was heated for one day under vacuum at 220°C. The resulting dry powder was pure NbO2F, with no impurity peaks detectable in the x-ray pattern. The starting materials were dried under vacuum at 220°C, weighed out in the appropriate ratio, mixed, and pressed into pellets. The pellets were loaded into platinum crucibles, heated at 400°C under vacuum for 1 hour and then sealed. The crucibles were then heated at 600°C for 2 hours to 1 day. Some mixtures were also heat-treated at 700°C for 1/2 day. Although stoichiometric NaNbO2F2 could be made in reasonably pure form, there was no evidence for the existence of the defect solid solutions, no matter what heat treatment was used. It appears that the hoped-for solid solutions do not form, and instead new phases appear. Efforts were made to 34 identify these new phases and for some compositions a tentative interpretation is given below. X-ray patterns for the Mo and W-substituted phases give a two-phase mixture of NaNbO2F2 and substantial amounts of an unknown phase. The x-ray patterns for Na0.7Y0.iNbO2F2, Na0.7Nd0>1NbO2F2 and Nao.8Cao.iNbO2F2 bear a strong resemblance to each other and may belong to the same structure type. These patterns could not be identified with any known Na-Nb compounds. However, a study of various sodium tantalum oxyfluorides has been published recently'12' and from the given cell constants for these tantalum oxyfluorides we calculated x-ray powder patterns. We found that the major x-ray peaks of Na0.7Y0.iNbO2F2, Na0.7Nd0>1NbO2F2 and Na0.8Ca0>1NbO2F2 can be indexed on the basis of the Na2Ta3O6F5-type unit cell. Similarly, the patterns of Nao.8Sr0.iNbO2F2 and Nao.gNbC^Fi.g can be indexed for the most part on the basis of the Na^asOioFg-type unit cell. These two structure types are unknown and their x-ray patterns have not been published; therefore, the tentative identifications made above are speculative. The x-ray pattern for Nao.8Pb0.iNbO2F2 is different from the others but bears some resemblance to that of Na0.8Sr0.iNbO2F2. It seems that the NaNbO2F2 structure does not tolerate defect solid solutions and that new structure types appear if the Na/Nb ratio is lowered. 3. Evaluation The sodium and fluorine NMR spectrum in Nao.sSro.iNbOfr^ were mea- sured at room temperature and 186°C (Figs. 14a, b). The peak-to-peak line widths were broad (about 8 gauss) and no motional narrowing was observed over this temperature range. The ionic conductivity of the pure and doped NaNbO2F2 specimens could not be measured because they were too porous. The dielectric loss and AC conductivity results are given in Table VI. E. Phases with a Defect Perovskite Structure 1. Structure and Transport Model The well-known perovskite structure is illustrated in Fig. 15 for the idealized cubic form. Normally the perovskite structure would not be considered as a good candidate for high cationic conductivity. In this ABX3 structure, the large A-cations are 12-coordinated and form together with the X-anions cubic close-packed layers perpendicular to the cubic [111] directions. In this close-packed array, the B-cations occupy octahedral interstices. In stoichiometric ABX3 perovskites, such as SrTiO3 or KNbO3, the high packing efficiency and lack of open channels would rule out any high cationic 35 (a) (b) Fig. 14 (a) NMR spectrum of 23Na in Nao.8Sro.iNbO2F2 at 186°C. (b) Spectrum of 19F. 36 TABLE VI AC Conductivity at Sample Additive Resonance Peak Room Temp, (ohm"1 cm"1) C-12 b Yes 103Hz 2 x 1Q-9 AE< 1 kcal (must be 104 1 x 10"8 electron blocking with 105 4 x 10~8 p elect ^108) 106 1 x 1CT7 C-18 Y No 103 1.9 x 10'10 104 1. 5 X 10"9 105 2 x 10'8 106 3 x 10"7 C-19 Nd No 103 2 x 10-9 104 1.4x 10"8 105 9 x 10'8 106 4 x 10-7 C-20 Ca No 103 8 x 10"10 104 3 x 1CT9 105 3 x 10'8 106 3 x lO'7 C-21 Sr Yes 103 1. 3 x 1Q-8 AE ~8. 1 kcal mol'1 104 9 x 10"8 p 25° ~ 6 x 106 ohm-cm 105 3 x 10"7 106 8 X 10-7 C-22 Pb No 103 5 . 7 x l O - 1 0 104 2.5 x ID"9 105 1.5 x 10'8 106 4 x 1CT8 mobility. Even if the structure is made nonstoichiometric by creating vacancies in the A-site, one would not normally expect these large A-cations to move easily to the vacant sites because the constrictions which separate the A-sites are too narrow. However, these objections can be partly overcome when we consider non- stoichiometric variants of the known cubic perovskite Lao.5Li0>5TiO3 (ao = 3. 86lA)(13). Here Li1 apparently occupies (randomly with La3+) the 12-coordinated A-site with twelve very large Li-O distances of 2. 73A. It is very unusual for an intermediate size cation such Li1+ to occupy a 12- coordinated site with such very large Li-O separations. More typical Li-O distances are of the order 1.95A to 2. 30A. 37 yFig. 15 The ideal or cubic perovskite structure. It was hoped that vacancies can be induced in this perovskite to provide for cation motion. This could be done by substituting a stable pentavalent ion such as Nb5+ or Ta5+ for tetravalent Ti. We could thus envision a cubic perovskite of composition (La0.5Lio.3^.zP^^CTio.sTao.2)^*03, where the intro- duction of 20% vacancies into the A-site could result in high lithium ion mobility. For this study it is especially significant that a cubic perovskite of composition (La2/3d1/3)TiO3 can be very closely approached, (13-17) ^^ that here density measurements definitely indicate vacancies in the A-site. The greatest stumbling block toward high Li1 mobility in a perovskite such as (Lao.sLio. 3no 2) (Tio.aTao.2) O3 is perhaps the narrow constriction through which the Lr+ ion must move to go from one A-site to one of the six neighboring A-sites. (On the average, about one of these six A-sites would be vacant.) At its closest approach to the other ions, Li1+ will be surrounded by four square planar oxygens at a distance of 1. 93A and four square planar Ti4+ (or Ta5+) ions at a distance of 2. 73A. However, this short Li-O bond is still within the range of typical 4-coordinated Li-O distances: 1. 86A to 2. 05A according to the International Tables for X-ray Crystallography. The motion of Li1+ ions through this constriction would therefore be possible, but not easy. 38 The phase diagram for the ternary system Li2O-La2O3-TiO2 has not been established and of the binaries, only the system La2O3-TiO2 has been studied over the entire range. (*8) However, fragmentary data are available on the other two binary systems. Recently cubic perovskites were found in the system Y over a narrow composition range. 0. 20 < x < 0. 24. We have investigated the possibility of using such a cubic perovskite--e. g. , ao.78Yo.22(Ta0.66W0.34)O3- as a base for introducing small amounts of Li and Na in the 12 -coordinated site. 2. Synthesis For the titanium -perovskites, the following compositions were attempted: (La0.5Lio.3Po.2) (Tio.8Ta0.2)O3, (La0.5Nao.3Do.2) (Nd0t5Li0.3a0.2) (Ti0.8Tao.2)03, (Ndo.gNao.sOo.a) (Tio.8Tao.2)O3. Li2CO3, Na2CO3, Ta2O5, TiC>2, La2O3, and Nd2O3 were intimately mixed in the appropriate ratios. In the attempted synthesis, two approaches were used: 1. The mixes were loaded into platinum boats and heated at 1200°C for 2 hours. Then the samples were 'withdrawn from the furnace and cooled in air. 2. The mixtures were dissolved in concentrated nitric acid. The resulting solution was heated to dryness in platinum crucibles. The crucibles were then transferred to a muffle furnace and heated very gradually from 300° to 1200°C. The temperature was maintained at 1200°C for 2 hours and the crucibles were subsequently quenched by quickly immersing them in cold water. Neither of the two methods of synthesis resulted in the hoped for cubic defect perovskite. Instead, a birefringent product was obtained in all cases. The complex x-ray patterns of the products have not yet been interpreted completely, but the more intense x-ray peaks are near the peak positions expected for a cubic perovskite. It appears that the products may be ordered, distorted perovskites. For the tantalum -tungsten perovskites, the oxides were mixed as shown in the equation below: 0. 05 Li2WO4 + 0. 099 Y2O3 + 0. 347 Ta2O5 + 0. 256 WO3 = Lio.ioYo.i98 (Tao.694W0.3o6)O3. 39 To make the Na-analogue, Na2WO4 • 2H2O was substituted for Li2WO4. These two mixtvires were pressed into pellets and reacted for one-half hour at 1200°C. In a modification of the above procedure, the samples were sealed into platinum tubing, heated at 1400°C for about one hour, and then quickly withdrawn and quenched in cold water. None of the resulting products appeared to be of the cubic perovskite type. All samples were birefringent with a more complex perovskite-related x-ray pattern. It appears that for all the attempted perovskite s, the undistorted cubic structure could not be obtained. The distorted (and ordered) perovskites which were apparently formed would not be considered good candidates for cationic conductivity since the structural perovskite-like framework has probably collapsed around the small Li1+ or Na1+ sites, making ionic mobility of these ions less likely. 3. Evaluation The two Ta-W perovskite-type samples were evaluated by dielectric loss. No loss peaks were observed and it is unlikely that these materials are good ionic conductors at low temperature. As pointed out in Section 2, we attribute the absence of conductivity to the collapse of the perovskite-like framework around the small monovalent cations. F. Sodium Tungsten Bronze-type Compounds 1. Structure and Transport Model A considerable effort was made last year in attempts to determine the ionic transport in nonstoichiometric lithium vanadates which are mixed electronic-ionic conductors. Since the differentiation of electron and ion charge carriers is a problem requiring investigation in all solid conductors, it was desirable to observe the nature of the response of our newly developed fast-pulse technique to materials of that kind. Sodium tungsten bronze was chosen for the test instead of lithium vanadate because large single crystals were available. The Nao. gWO3 structure is essentially a slightly distorted version of the ideal (cubic) perovskite structure shown in Fig. 15. 10$ of the Na sites are vacant, and some degree of ionic motion might therefore be possible. Such ionic motion would be difficult to detect by conventional conductivity measure- ments, due to the very high electronic conduction. 2. Sample Preparation An electrolytically grown single crystal of Na0.gWO3 was obtained from Dr. R. C. DeVries of the General Electric Company. From a larger single 40 crystal, a smaller piece was cut for the cell measurements described below. 3. Evaluation The sodium tungsten bronze crystal was cut parallel to one of the pseudo- cubic crystal axes and mounted in a microcell. The conducting surface was a circular area 1. 0 mm in diameter and the crystal was 1. 2 mm thick. The response of the cell to pulses of long and short duration is shown in Fig. 16. A large current flow is observed, and in addition, a slow, capacitive-like charge and decay extending for approximately 20 |asec. The capacitive - like behavior was not observed when platinum was substituted for the sodium tungsten bronze and, therefore, it is unlikely the slow processes are due to polarization at the interface. The observations suggest there may be a threshold for current flow, possibly related to the displacement of sodium ions in the tunnels. G. NaLiGe4Og-Type Compounds 1. Structure and Transport Model The structure of the orthorhombic NaLiGe^g'2^ is illustrated in Fig. 17 and consists of (GeO3^n~ chains made up of GeC>4 tetrahedra sharing corners. These chains are connected by GeOe octahedra to form a three-dimensional framework. This framework gives rise to interwoven tunnels which are occupied by Na+ and Li"1" ions. The tunnels might be sufficiently wide to permit some alkali ion motion in three dimensions. The most promising ionic conductors are perhaps the nonstoichiometric variants where vacancies in the alkali ion positions are created by substituting pentavalent ions for Ge4+, either in the octahedral site (Nb , Ta5+, Sb5*) or tetrahedral site (V ). However, some ionic motion may take place even in the stoichiometric Li2Ge4O9 along interstitial sites. In their regular sites, Li1 ions are 5-coordinated with Li-O bonds at 1. 827A to 2. 397A at the corners of a square pyramid. Actually this Li1+ ion is distributed randomly over two equivalent sites 0. 6A apart, and there is a good chance that dynamic exchange between these two places may occur, which would enhance the likelihood of Li1 ion mobility along the channels. Na1+ ions occupy six-coordinated sites in different channels than the Li1+ ions, with Na-O distances ranging from 2. 326A to 2. 900A. This Na1+ ion can be completely replaced with Li1+ to form the end member Li2Ge4Og. If the Li"*" ions were to occupy exactly the same site as Na ions, there would be six very large Li-O distances (2. 55A on the average). It has been postulated^) (but not proved) that Li1+ would prefer a different site (removed by I/ 2 b0 = 2. 31A) along the same channel which would yield six shorter Li-O distances (2. 26A), on the average- Since Li1 ions have a low scattering 41 ••••••IBM •MHHII A'VMWM Fig. 16 Voltage response of sodium tungsten bronze crystal to constant current pulse. Above, 250 [jsec pulse; below, 20 usec pulse. 42 c - <£-, Fig. 17 The framework of interlinked GeO4 tetrahedra and GeO6 octahedra in the NaLiGe4O9 structure. After Ref. 23. factor and cannot be easily placed by x-ray diffraction study, there is some uncertainty regarding the sites occupied by Li1 ions in Li2Ge4O9. However, the intensity agreement is a little better if Li1 ions are placed in the site giving shorter Li-O bonds. It seems possible, however, that both sites could be partially occupied, a factor that might enhance ionic mobility. 2. Synthesis A phase equilibrium diagram is available for the system Li2O-GeO2, however, it is partly incorrect in the region of interest. ^2) Since glasses can be quenched from the Li2O-GeO2 melt, it was decided to synthesize and examine both glassy and crystalline samples. Several glasses near the Li2Ge4O9 composition have been prepared: Li2Ge4O9, NaLiGe4O9, Li^Ges^Nbo.sOg, ^^Ge^Tao^OQ, Lii.7Ge3.7Sbo.3Og and Lii^Ges^Vo.sOg. To synthesize these glasses, the appropriate amounts of GeO2 andLi2CO3 (or Na2CO3) were intimately mixed and heated in air at 900°C in a platinum crucible. The resulting powder was then mixed with appropriate amounts of the other oxides (Nb2O5, Ta2O5, Sb2O5 and V2O5). 43 The powders were loaded into platinum capsules and sealed. The sample capsules were heated at 1200°C for 1/2 hour and then quenched by quickly immersing tnem into cold water. The quench rate must be quite rapid or partial crystallization occurs. If larger mixtures of powder are heated in a platinum crucible, a clear glass cannot be quenched from 1200°C due to partial crystallization on cooling. In such cases clear glasses could be recovered only as a skin near the platinum wall. The glasses were homogeneous, clear, and colorless, except for the V -substituted glass which was amber-colored. The glasses were checked microscopically for homogeneity by observing the refractive index. Initially some problems were encountered in the attempted synthesis of crystalline Li2Ge4Og and substituted variants. When the starting materials were heated at various temperatures below 1200°C, the resulting Li2Ge4O9 phase was impure and x-ray analysis indicated the presence of significant amounts of Li2GeO3, Li2Ge7O15 and GeO2. Purer crystalline products could be obtained by heating the glasses, described above, in sealed platinum tubes, at 650° to 700°C for several days. The crystalline products thus obtained gave the expected x-ray pattern for stoichiometric Li2Ge4O9 free from impurity peaks. For compositions Lii.7Ge3.7Tao.3O9 and Lii.TGea^Nbo.sOg impurities of the Li2Ge7O15 phase (5$ to 15$) were detected. For the composition NaLiGe4Og a pure crystalline material could be prepared by either of two methods: (a) Heating NaLiGe4O9 glass at 700°C for several days. This results in a very finely crystallized sample. (b) Heating the appropriate amounts of Na2CO3 and GeO2 slowly in a platinum crucible to 1200°C, and then withdrawing the crucible and letting it cool in air. This gives a coarsely crystallized material. 3. Evaluation The compositions Li2Ge4O9 and NaLiGe4O9 were prepared as glasses and screened by NMR for cation mobility at room temperature. The glasses were then devitrified and NMR screening repeated on the crystalline phases. The NMR results are summarized in Table VII. There is strong motional narrowing of the lithium resonance in the glass form of Li2Ge4O9. The peak width at room temperature (Fig. 18) is only 1. 4 gauss and it narrows further to 0. 4 gauss at 153°C. The lithium spectrum in the crystallized compound is nearly the same at room temperature (Fig. 19), but at 115°C the spectrum is resolved into two peaks: one with width 1. 5 gauss, the second 0. 35 gauss. This agrees with the prediction made on the basis of structural considerations that lithium atoms might occupy two different crystallographic sites. The latter motionally narrowed peak is probably due to lithium ions with large Li-O distances which have replaced the sodium in the 6-fold sites occupied by Na in NaLiGe4O9. 44 TABLE VE NMR Spectra of 7Li in Lj2Ge4Og Compounds Material Li2Ge4Og Phase Line Width Temp (°C) (Gauss) Glass Li2Ge4Og Polycrystal NaLiGe4O9 Glass Polycrystal -100 22 120 163 22 115 -34 22 95 22 115 r 1 i r-l- 'i-; I i I I | I I i • [ r r r-r J I T- • I [ rr T t [ i |-|' • ;-i 1.9 1.4 0. 8 0.4 1.5 0. 35 and 1. 5 1. 6 1. 6 1. 6 1.3 1. 3 1 1 ' ' ' ' I ' ' ' ' | ' ' ' ' I ' ' ' ' I ' ' ' ' I ' ' ' I I ' •wa . I . . . . ...... , , , . 1 7T ^ ,-.Fig. 18 NMR spectrum of Li in glass with composition Li2Ge4Og at room temperature. The peak width is 1. 4 gauss. 45 Fig. 19 NMR spectrum of 7Li in crystalline Li2Ge4O9 at room temperature. The peak width is 1. 5 gauss. Motional narrowing of the lithium NMR spectrum was not observed either in the glass or in the crystalline phases of NaLiGe4O9 (Figs. 20 and 21). This i6 consistent with the structural interpretation suggested to explain the Li2Ge4O9 spectra because sodium ions are significantly larger than lithium ions. The sodium spectrum could not be detected despite prolonged signal averaging (25 hours). This indicates the spectrum is broadened by the crystalline fields, and the sodium ions are relatively immobile. The pulse shape in one specimen of Li2Ge4O9 glass was successfully observed in a microcell. Attempts to measure the other compositions failed because the materials cracked and developed leaks when they were helium tested. The glass was 0. 24 mm thick with an effective cross section of 2. 5 x 3. 5 mm. The cell response indicated the conductivity is small, in agreement with dielectric loss measurements made at NASA-Lewis Research Center. No resonance peaks were observed and the room-temperature AC conductivity was 4 x 10"10 at 103 Hz and 2 x 10'7 ohm'1 cm"1 at 106 Hz. H. L,izGe7Oi5-Type Compounds 1. Structure and Transport Model The structure of Li2Ge7O15(23) is related to the NaLiGe4O9 type discussed above. In Li2Ge7O15, GeO4 tetrahedra are linked together at the corners to form corrugated layers which are connected by GeO6 octahedra to form a 46 I I I I I I I I I • . I I I I I I I I I • I ' ' ' ' I ' ' ' ' I ' ' ' ' I ' '•' ' I ' ' ' ' I ' ' ' ' I ' Fig. 20 NMR spectrum of 7Li in glass with composition NaLiGe^g at room temperature. The peak width is 1. 6 gauss. Fig. 21 NMR spectrum of 7Li in crystalline NaLiGe4O9 at room temperature. The peak width is 1. 3 gauss. 47 loose three-dimensional framework. This can be seen from Fig. 22. Within this framework there are interwoven channels which can accommodate Li1+ ions. (2^ The two Li1+ sites both have 6-coordinated Li1+ ions with o o Li-O distances varying from 1. 933A to 2. 367A. The structure is so loose that it might accommodate larger ions (such as Na ) in interstitial sites. One such 6-coordinated interstitial site has an average X-O distance of 2. 563A. (23) Three-dimensional motion of alkali ions is conceivable both for stoichiometric Li2Ge7O15 and nonstoichiometric variants (described below), since there are interstitial sites and wide enough tunnels in this loose framework structure. The low density and " looseness" of the Ld2Ge7O15 structure is apparent from the large molecular volume (298. 8A3) when compared to the molecular volume of 218. 2A3 derived from the component oxides Li2O and tetragonal GeO2. b =16,76 A Li2Ge7O15 c-3.69A1* V Fig. 22 The structure of Li2Ge7O15 projected on the (100) plane. 48 Nonstoichiometric L^GeyOj^ type phases can be envisioned by substituting pentavalent ions for Ge4 and creating vacancies in the Li1+ sites, e.g., Li2-xGe7_xNbxOi5. Alternately one could try to force additional alkali ions (like Na1+) into the large interstitial holes by making solid solutions of the type Li2NaxGe7 -xAlxOi5. 2. Synthesis Both glassy and crystalline phases have been prepared for Li2Ge7O15 and substituted derivatives. (a) Li2Ge7Qi5 Larger amounts of Li2Ge7O15 (20 to 50 grams) have been prepared by slowly heating an intimate mixture of the appropriate amounts of Li2CO3 and GeO2 from 450° to 1000°C in a platinum crucible, holding the temperature at 1000°C for one hour, and then withdrawing the crucible from the furnace. The Li2Ge7O15 product so obtained is a very fine crystalline powder of high purity. X-ray patterns show either no detectable impurity peaks or only traces of GeO2. When the above -described synthesis is carried out at significantly higher or lower final heating temperatures, pure Li2Ge7O15 is not obtained. A synthesis at 112 0°C resulted in large amounts of excess GeC>2. A low- temperature preparation (850°C) contained in addition to Li2Ge7Oi5 significant amounts of four other phases: GeO2 (quartz structure), GeC>2 (rutile structure), 9, and Li2GeOs. Glassy Li2Ge7O15 can be made by heating the appropriate Li2CO3-GeO2 mixture (or crystalline Li2Ge7O15) slowly up to 1200°C in a platinum tube or crucible. The platinum containers are then quickly withdrawn from the furnace and immersed in cold water. If a slower cooling rate is used, partial crystallization takes place. The Li2Ge7O15 glass has a refractive index of 1. 690. (b) Substituted glasses near the Li2Ge7Oi5 stoichiometry Homogeneous glasses of the following compositions were made and are listed below with their refractive indices: Composition of Glass Refractive Index Lilt7Ge6.7Sbo.3Pi5 1. 692 LiijGee^Tao.aOis 1. 698 Lii.7Gee.7Nbo.3Oi5 1. 698 Lii.7Ge6.7V0.3015 1. 688 1. 676 49 The glasses can be made by heating the appropriate mixtures of crystalline Li2Ge7O15, GeO2, Sb2O5, Ta2O5, Nb2O5, and V2O5 in sealed platinum tubes for 1/2 hour at 1200°C, and then rapidly quenching the tubes in cold water. The Lii.TGeg^Tao.aOis glass has the lowest tendency to crystallize, and this composition can therefore be prepared in larger batches by slowly heating the appropriate mixture of oxides in an open platinum crucible to 1200°C, and then removing the crucible from the furnace. The L^Nao.aGee^Alo.aOjs glass can be made by heating the appropriate mixture of Li2CO3, Na2CO3, GeO2, and A12O3 at 850°C for two hours. The resulting mixture was then sealed into platinum tubes and heated for 1/2 hour at 1200°C. The tubes were then quenched in cold water. (c) Crystalline, Nonstoichiometric Li2Ge7O15-Type Phases Efforts were made to prepare nonstoichiometric Li2Ge7Oi5-type phases by heating the appropriate mixtures of oxides and LigCC^ from 450° to 1000°C. This method, however, yielded pure samples only for stoichiometric Li2Ge7Oi5. Even when the heat treatment was modified, nonstoichiometric Li2Ge7O15 type phases could not be prepared in the pure state. A successful synthesis of pure crystalline Li1>7Ge6>7Tao.3O15 and Lii^Gee.fSbo.sOis was finally achieved when glasses of the same compositions were used as starting materials. The glasses were annealed at 600°C for about three days. The x-ray patterns of the products showed only peaks belonging to the Li2Ge7O15 structure, indicating a high phase purity. Attempts to anneal the Li2Na0.3Ge6t7Alo.3O15 glass at 600°C did not result in a pure crystalline phase. While the major phase was of the Li2Ge7O15 type, smaller (but significant) amounts of NaLiGe^g and a quartz type phase were also present. 3. Evaluation Strong motional narrowing of the 7Li NMR spectrum is seen in both Li2Ge7O15 and Lii.7Ge6.7Tao>3O15 glasses (Figs. 23, 24, and Table VIII). Devitrifying the glasses resulted in slight line broadening indicating that crystallization tends to immobilize the lithium atoms. The lithium spectrum in the crystalline compounds is still narrow (1. 0 gauss) and the critical temperature for motional narrowing may have been displaced to a higher value. The conductivity in Li2Ge7O15 glass and in the tantalum doped glass was measured in a microcell. The response is nearly capacitive (Fig. 25), but there was a slight current flow indicating small conductivity. 50 . I . . I . I . . . . I . . . . I . . . . I . . . I . . I I I . . I.— , , . , I . . . . I . . . . I . . . . I . Fig. 23 NMR spectrum of 7Li in glass with composition Li2Ge7O15 at room temperature. The peak width is 0. 9 gauss. Fig. 24 NMR spectrum of 7Li in crystalline Li2Ge7O15. The peak width is 1. 0 gauss. 51 TABLE VIII NMR Spectra of 7Li in LiaGe?PIS-Type Compounds Material Li2Ge7O15 Lii.7Ge6.7Tao.aOi5 Phase Glass Crystal Glass Crystal Temp (°C) 22 165 22 170 -34 22 96 22 95 Line Width (Gauss) 0. 9 0.45 1.0 1.0 1.2 1.0 0.65 1.0 1.0 The dielectric loss and room temperature AC conductivity measure- ments on glassy and crystalline L1izG&jOi5 compositions are summarized in Table IX. TABLE IX AC Conductivity at Room Temp Sample Phase (ohm"1 cm"1) Li2Ge7Ol5 Glass Polycrystal Lii.7Ge6>7Tao.3Pi5 Glass Polycrystal 103Hz 105 103 10s 103 • 105 103 104 105 106 ~io-6 ~io-6 <5 x ID'12 ~1.5 x 10-9 ~5 x 10 -10 3 x 10"8 1.5 x 10-10 1.4 x 10"9 1.2 x 10"8 <5 x 10-8 52 Fig. 25 Voltage response of Li2Ge7O15 glass to constant current pulse in microcell. A small current flow can be seen in the expanded region of the trailing edge of the pulse. 53 The dielectric loss showed no peaks, but a small AC conductivity was measured at room temperature in lithium germanate glass. The reason for the smaller AC conductivity in the doped glass is not understood. In summary, there is strong short-range motion of lithium atoms in pure and doped I^GevOis glass, and this motion is decreased when the phase is crystallized. The ionic conductivity due to long-range movement of lithium ions is small but possibly was detected in Li2Ge7O15 glass. I. Hexagonal NaLnF4-Type Compounds 1. Structure and Transport Model The hexagonal NaNdF4 structure is the stable low-temperature form for all NaLnF4 phases (Ln = lanthanon or yttrium). For many NaLnF4 com- pounds a phase transition takes place at higher temperatures to the well- known cubic fluorite structure. (24-26) The structure of the hexagonal NaNdF4^24^ is shown in Fig. 26. There are three crystallographically independent sites for the cations. One is a 9-coordinated site and occupied only by Nd ions. Another 9-coordinated 0.5 No, 0.5 Nd 0.5 No O F Fig. 26 The structure of hexagonal NaNdF4. 54 site has 50$ Nd3+ and 50$ Na1+ ions distributed at random. A third octahedral site is half empty and half occupied by Na1+ ions. The structure study ^ 4) shows that the Na* ions in these half -empty Na -sites have strong anisotropic thermal motion, as do some of the fluoride ions. The thermal parameters of the Na1+ ion may reflect the uncertainty of site occupancy rather than true thermal motion. Our interest in the NaNdF4 -com pounds was primarily in the high- temperature cubic fluorite type solid solution phases {25, 26) rather than in the low -temperature hexagonal structure. (24) jj- was recognized, however, that the high-temperature fluorite form was more likely to be a F~ ion con- ductor than a Na1+ ion conductor. Since there was no interest in potential F~ ion conductors (for practical reasons) the earlier preliminary studies on the cubic NaLnF4 phases (2) were dropped. The studies on the hexagonal phases were continued since high ion mobility had been established in the cubic form, and it was possible the Na ion in the half -empty site might be mobile. 2. Synthesis of NaNdF4-Type Phases Hexagonal NaPrF4, NaNdF4, and NaYF4 were prepared as follows : commercially obtained NaF, PrF3, NdF3, and YF3 were dried at 220°C under vacuum for about a day. The starting materials were then mixed in a 1: 1 molar ratio and compressed into pellets. The pellets were placed in a platinum crucible and heated at 400°C under high vacuum to eliminate any traces of moisture. The crucible'was then sealed and heated at 750°C for 7 hours and subsequently at 600°C for 12 hours. The pellets were analyzed by x-ray diffraction, which showed that the hexagonal NaYF4 and NaNdF4 phases were free from detectable impurities. In the NaPrF4 sample small amounts of NaF were detectable. No attempt was made to prepare nonstoichiometric phases of the hexagonal NaNdF4 type. It is well known from a study of the phase equilibria of the NaF-LnF3 systems (25, 26) tnat these hexagonal phases do not form extensive solid solutions, unlike the high -temperature cubic forms. A sample of NaYF4 was hot pressed for 20 minutes at 7700 psi and 620° to 630°C. The specimen was contained in a graphite die and shown by x-ray diffraction to still be of the hexagonal modification, with no impurity peaks detectable. 3. Evaluation Three samples of hexagonal NaLnF4 compounds were examined for evidence of motional narrowing using wide -line NMR (Table X). Both the 23Na and 19F spectra were run at different temperatures. The room- temperature spectra of NaYF4, NaPrF4, and NaNdF4 are shown in Figs. 27 through 32. The peak-to-peak line widths measured at various temperatures 55 TABLE X NMR Spectra in Hexagonal NaLnF4 Compounds Material Nucleus NaYF4 Na F NaPrF4 Na F NaNdF4 Na F Temp (°C) 22 90 186 22 186 22 90 186 22 186 22 90 186 22 186 Width 6.0 6.0 6.0 9.'0 10.0 8.0 8.0 8.0 7, 5, 8.0 8.0 8.0 7, 5, (Gauss) 30 19 34 20 are given in Table X. There is no evidence for motional narrowing of the sodium nuclei in any of the samples. On the other hand, some of the fluorine spectra did show some decline in line width with increasing temperature. The fluorine spectra are complex, apparently having a narrow line superposed on a broad line. The mechanism of the broadening is not understood. 19F has no quadrupole moment and any broadening effects would arise from its dipole moment. The line width of the broad lines is believed too great to be accounted for on the basis of coupling between fluorine neighbors and may arise from coupling between the fluorine and the rare earth ions. This is consistent with the less broad line observed for the yttrium compound. The absence of motional narrowing of the sodium nuclei is in contrast to what was observed in the cubic NaLnF4 compounds last year. Motional narrowing of both the sodium and fluorine spectra was observed and the results could be interpreted by motion of either or both kinds of nuclei. 56 23TFig. 2 7 NMR spectrum of Na in hexagonal NaYF4 at room temperature. The peak-to-peak width is 6. 0 gauss. Fig. 28 NMR spectrum of 19F in hexagonal NaYF4 at 186°C. The peak width is 10 gauss, approximately the same as room temperature. 57 . . . . . . . . . . . - , . . . . ? . . . . , . . . . , . . . . , . . . . , . . , . , . . . . , . . . . , . . . . , . . . . . . . V Fig. 29 NMR spectrum of 23Na in hexagonal NaPrF4 at room temperature. The peak width is 8. 0 gauss. Fig. 30 NMR spectrum of 19F in hexagonal NaPrF4 at room temperature. The spectrum is complex and apparently consists of a narrow peak superposed on a broad peak. 58 1 I ' - • • I ' ' ' ! 1 '—' - I ' ' ' ' I ' • ' ' I • -—I ' ' ' ' 1 ' ' 1 • ' ' ' I • I ' ' ' ' I ' ' ' ' I ' ' • ' I ' ' ' ' I ' ' ' ' I ' ' ' ' I ' ' ' ' I ' ' ' ' I ' I . . . . I . , . , I . . . . I . I . . . . I . . . . I . . . . I I . . . . I . . . . I . . . . I . Fig. 31 NMR spectrum of 23Na in hexagonal NaNdF4 at room temperature. The peak width is 8. 0 gauss. Fig. 32 NMR spectrum of 19F in hexagonal NaNdF4 at room temperature. The two fluorine peaks are resolved. 59 Since narrowing of the fluorine spectrum alone is seen in the hexagonal com- pounds, the sharpening of the sodium spectrum in the cubic compounds was probably caused by sodium motion and was not due to motion of the neighboring fluorine nuclei. Dielectric loss and room temperature AC conductivity measurements made at NASA-Lewis Research Center indicated possible ionic conductivity in hexagonal NaNdF4 and NaYF4. Their results are given in Table XI. Sample C-15a NaPrF4 TABLE XI _ Resonance Peak _ Yes (two peaks observed: AE = 5. 4-6. 9 kcal from low- temp peak may be Na motion; p 25°C ~1 ohm-cm AE = 15. 9 kcal from high- temp peak may be F motion); p25°C = 5 x ! 0 5 AC Conductivity at Room Temp (ohm"1 cm" ) C-16 NaNdF4 C-17 NaYF4 C-17as NaYF4 hot-pressed (106 Hz, Yes 125°C, Cole-Cole conductivity ~2 x 10~6 ohm"1 cm'1 at 125°K) (106 Hz, Yes 12 5° K, Cole-Cole conductivity ~5x 10~6 cm-1 at 125°K) Yes AE = 4. 3 kcal p 25°C ~ 10 ohm-cm ohm _i 10d Hz 104 105 106 103 104 105 106 103 104 105 106 103 104 105 106 5 x 10~9 6 x 10 -8 1. 4. 1. 6. 2. 6. 2. 1. 4. 2. 1. 1. 5 2 6 6 4 6 9 4 1 8 6 4 9 x x X X X X X X X X X X X X 10 10 10 10 10 10 10 10 10 10 _7 -6 -8 -8 -7 _7 -8 -8 -7 -6 io-q 10 10 -8 -7 io-H We conclude from the dielectric loss and AC conductivity measurements that the hexagonal NaLnF4 compounds are moderate conductors at low temper - ature. The conductivity probably could be increased by doping, but the NMR spectra show the phase is an anion conductor and consequently of little interest in the present context. J. Dalyite-Type Compounds 1. Structure and Transport Model The structure of the rare mineral dalyite (ideally K2ZrSi^O15) has recently been determined by Fleet. (27) The structure is illustrated in Figs. 33 and 34, and can be considered as a layer structure of a new type. The layers are 60 Pig. 33 The dalyite structure (ideally K^rS^C^*,): o = Si; o = 0- O = Zr- O = K. After Ref. 27. ' made up of SiO4 tetrahedra sharing corners to give four, six and 8-membered rings, as illustrated in Fig. 33. These silica sheets are held together in the third dimension by ZrO6 octahedra, giving rise to a three-dimensional frame- work with interconnected channels, occupied by K+ ions. Motion of the K+ ions seems possible in this structure, particularly if we consider the large temperature factor for K+ (B = 3. 8A3 in the a-direction). The K-O distances are also larger than average (2. 83A to 3. 49A) which may indicate a fairly loose K ion. 61 an) Fig. 34 Idealized dalyite sheets made up of SiO4 tetrahedra sharing corners. After Ref. 27. The structure would be even a more attractive candidate if nonstoichiometric variants could be made and if the K+ ion could be exchanged with Na+ or Ag1 . To make the structure nonstoichiometric, one could substitute Zr4+ with Ta5+, i. e., K+ + Zr4+ = D + Ta5+. It might also be possible to exchange K+ with Na+ or Ag1+ in a nitrate melt. It is well known that, in the mineral dalyite, significant amounts of Na+ (~15$) can replace the K1+ ions, '^7, 28) and the smaller Na+ or Ag' structure. ions might be very loose and mobile in the 2. Synthesis of Dalyite-Type Phases So far, synthetic dalyite has been prepared only under hydrothermal conditions at low temperatures and efforts to make the phase by more conventional sintering techniques have been unsuccessful. (29, 30) it was felt that an effort should be made to synthesize K2ZtrSi6O15 by a more easily accessible method. 62 Reagent grade K2CO3, ZrOSO4- H2SO4- 3H2O, and silicic acid (SiO2- xH2O) were intimately mixed, pelletized, and heated gradually up to 1200°C in a platinum container. At this stage the resulting product did not contain any dalyite-like phase but was a mixture of at least four different products, as observed by optical and x-ray methods: a glassy phase, a cristobalite-like crystalline material, ZrC^, and an unidentified crystalline product. This mixture was now heated for 20 minutes at 1700°C in a platinum boat, and cooled in air. The resulting product was a homogeneous glass. (Attempts to prepare such glasses at lower temperatures were unsuccessful.) When this glass was heated for one day at 1200°C, the resulting product had completely crystallized. The x-ray pattern of this crystallized product is not identical to the published dalyite patterns, '^8, 29) ^ ^ y. nevertheless bears a resemblance to these published patterns. It is possible to interpret these data as indicating the presence of a dalyite-related phase, not identical to the mineral dalyite, but similar to it. Since the dalyite work was initiated two weeks before the termination of the contract, no further synthesis could be carried out. 3. Evaluation The " dalyite11 samples could not be synthesized in time for evaluation before the expiration of the contract. V. CONCLUSIONS AND RECOMMENDATIONS A large number of structure types and many compounds have now been evaluated as possible solid electrolytes for high-energy and power-density batteries to operate at ambient temperature. A compilation of the results obtained this year is given in Table XII. The crystal-chemical basis for predicting potential conductors has been validated in many cases. We have obtained evidence for ionic conductivity in one-dimensional and three- dimensional channel structures. No new solid electrolyte has yet been dis- covered with the requisite properties of high cation conductivity and chemical stability for application in a practical low-temperature battery, but we believe further work as outlined below is warranted and has a high probability for success. We have obtained new information which allows us to make the crystal-chemical criteria for high ionic conductivity more restrictive. In addition, we have developed significant new methods for evaluation of ion transport. In this section we have summarized the conclusions which result from the present study and make recommendations for future work. 63 N hriHM ^ .tf 8 -^ •s S « rt «•Hi o U CO O COI ij m CQ CQ 0 tf bfl « 0 0 JH (P Cfl U O CQ 0 O 0 I—I Si0 -t-> rt rt S CO CQ CQ -§ CO rt 60) 0 HH rt M ^J¥ cog rt" . U SH rt J rt CQ r3 rt fcuoH CQ CO 15 k" OH 0 O CO c? 1 •H O fe h O U CM ft O a > o o^> tuo ff) c- a> O rt •* fe a rt !? ffi 13a .2 w a0 P r— 1 rt fl O •H CQ C 0 g .3p CM i — 1 rtflS •rH £j 0 P 1 co 64 A. Conclusions The principal conclusions resulting from this study are as follows: 1. Crystal chemistry has been used successfully to predict and identify many structures in which cations are loosely bound and capable of at least short-range motion at low temperature. 2. Compounds with this degree of ionic mobility can be screened by NMR and dielectric loss. 3. Short-range motion is a necessary although not sufficient condition for ion transport. In the compounds that have been evaluated thus far the long-range ionic motion or transport has been small at room temperature. 4. Our investigation of channel structures has led to an important refinement of the crystal chemical criteria for the selection of candidate cation conductors: (a) The channels through which ion migration is expected to occur should be interconnected in at least two dimensions, and preferably in three dimensions. (b) Each potentially mobile ion should have available a minimum of three vacant sites into which it can move in a single elementary jump. 5. We have developed and validated a new fast pulse technique for the measurement of ion transport in very small solid specimens. This method is extremely promising and we believe it probably will prove to be a major breakthrough in evaluating ion transport in solids. The method will be particularly useful for measuring materials that are difficult to prepare as large single crystals or as pure dense poly crystalline solids. 6. Microwave measurements can be used to measure the conductivity of powders. B. Recommendations We believe that work in the future should be restricted to three- and two-dimensional interconnected channel structures, and that evaluation should be made principally by the fast pulse technique. Dielectric loss and/or microwave conductivity on powders are useful adjunct measurements. 65 A broad family of anhydrous zeolitic*-like structures related to analcite (NaAlSi2O6) has been selected for a focused search. The following considerations were used to arrive at this class of compounds: (1) The channels should be interconnected in three-dimensions to avoid blocking; (2) The channel diameter should be slightly larger than the dimensions of nonhydrated ions; (3) It should be possible to modify the materials in a relatively continuous manner over a broad range of composition and structure; (4) It should be possible to fabricate the materials into dense nonporous membranes for battery applications; and (5) There should be x-ray or other evidence of ion mobility in at least some of the compositions. The starting points chosen for our zeolite system are the anhydrous compositions NaAlSi2O6 (analcite), KAlSi2O6 (high leucite), and CsAlSi2O6 (pollucite). These are well-known minerals and their structures have been described. (34-39) All three of these minerals are cubic with space group Ojf-IaSd and with cell constants around 13. 7A. The zeolitic, three-dimensional alumino- silicate network is the same for all three minerals, but only high-leucite and pollucite are truly isotypic. The A1C>4 and SiO4 tetrahedra are randomly interlinked at all four corners to form an open three-dimensional framework. *It is appropriate to comment on two objections usually encountered when zeolites are considered as potential solid electrolytes. The first of these involves the fact that it is well known that removal of water from the zeolite structure results in disappointingly low conductivities and diffusion rates. This is a. valid concern, but there have not been intensive studies of ion transport in nonstoichiometric, anhydrous, small channel zeolites of the type considered for this study. In fact, the channels in the proposed com- pounds are so small it may be somewhat inaccurate to refer to them as zeolites. A second objection is that since most zeolites are porous and composed of extremely small crystallites, it may be difficult to fabricate into dense non-porous membranes for practical applications. We believe this is not necessarily a valid objection since some anhydrous zeolites can be made by devitrifying glasses of the same composition. For example, KAlSi2O6 (leucite), and CsBSi2O6 (pollucite structure) can be made by devitrifying the appropriate glasses. (32, 33) we feel that if glasses can be used as precursors to the crystalline materials, it may be possible to fabricate dense membranes. 66 This leaves large channels, which in high leucite and pollucite, are occupied by potassium and cesium ions, respectively. The potassium and cesium ions are surrounded loosely by 12 oxygens at K-O distances of 3. 35A to 3. 54A and Cs-O distances of 3. 40A to 3. 57A. In hydrated specimens, water can partly occupy these 12-coordinated sites (or adjacent sites in pollucite). The analcite structure differs from that of pollucite and high-leucite in that sodium ions occupy the " windows" that connect the large cavities. The Na-O distances are approximately 2. 4A. These smaller window sites are more suited for the smaller sodium ions than are the larger cavities. We believe high ionic conductivity might be obtained in nonstoichiometric zeolites by the creation of additional vacancies (by adjustment of the Si/Al ratio) and by the simultaneous substitution of potassium ions for sodium or cesium ions. These intermediate-size ions may be small enough to move through the " windows, " yet small enough to be only loosely bonded inside the large cavity. Such loose bonding has been observed in high temperature leucite, KAlSi2O6. ^36' Here, the ion inside the large cavity is bonded to six oxygens at 3. 37A and to six oxygens at 3. 57A, with the potassium ions having the large isotropic temperature factor of 15. 4A~1. In such an environment, potassium ion mobility appears possible, since K-O distances normally average about 2. 7A to 2. 9A. Unfortunately, the high form of KAlSi2O6 cannot be quenched and reverts to a distorted low temperature from at about 605°C. By crystal-chemical substitutions, it may be possible to lower this transition temperature. VI. REFERENCES 1. Lundberg, M., Acta Chem. Scand. 19, 2274 (1965). 2. Roth, W. L., Mitoff, S. P., and King, R. N., Final Report, "Study, Selection and Preparation of Solid, Cationic Conductors, " Sept. 1972, Contract NAS3-15692, NASA-Lewis Research Center, Cleveland, Ohio. 2a. Rice, M. J. and Roth, W. L., J. Solid State Chem. 4, 294(1972). 2b. Kautz, H. E., Singer, J., Fielder, W. L., and Fordyce, J. S., NASA TN D-7146, " Electrical Screening Procedure for Solid Ionic Conductors, " March (1973). 2c. Roberts, S. and von Hippel, A., J. Appl. Phys. 17, 610 (1946). 3. Bystrbm, A. and Bystrbm, A.M., Acta Cryst. 3, 146 (1950) and Acta Cryst. 4, 469 (1951). 67 4. Bayer, G. and Hoffman, W., Am. Mineralogist 5J_, 511 (1966). 5. Kinomura, N., J. Am. Ceram. Soc. 56, 344 (1973). 6. Bystrbm, A., Arkiv Kemi Min. Geol. 15B [12], 1 (1942). 7. Gibbs, G. V., Am. Mineralogist 5JL, 1068(1966). 8. Smith, J. V. and Schreyer, W., Min. Mag. 33, 226(1962). 9. Schreyer, W. and Schairer, J. F., J. Petrology 2_, 324(1961). 10. Provondra, P. and Langer, K., Neues Jahrbuch fUr Mineralogie, Abhandlungen 116, 1 (1971). 11. Anderson, S. and Galy, J., Acta Cryst. B25, 847(1969). 12. Chaminade, J.-P., Pouchard, M., and Hagenmuller, P., Compt. Rend. 273C, 984 (1971). 13. Brous, J., Fankuchen, I., and Banks, E., Acta Cryst. 6, 67(1953). 14. Kestigian, M. and Ward, R., J. Am. Chem. Soc. 77, 6199 (1955). 15. Johnston, W. D. and Sestrich, D., J. Inorg. Nucl. Chem. 20, 32 (1961). 16. Tien, T. Y. and Hummel, F.A., Trans. Brit. Ceram. Soc. 66, 233 (1967). 17. Lejus, A.M. and Queyroux, F., Compt. Rend. 271C, 56 (1970). 18. MacChesney, J. B. and Sauer, H. A., J. Am. Ceram. Soc. 45, 419 (1962). 19. Studer, F., Montfort, Y., and Raveau, B., J. Solid State Chem. 7, 269 (1973). 20. Vollenkle, H., Wittmann, A., and Nowotny, H., Mh. Chem. 100, 79 (1969). 21. Krishna Murthy, M. and Ip, J., J. Am. Ceram. Soc. 47, 328 (1964). 22. Wittmann, A. and Modern, E., Mh. Chem. 96, 581 (1965). 23. Vollenkle, H., Wittmann, A., and Nowotny, H., Mh. Chem. 101, . 46 (1970). 68 24. Burns, J. H., Inorg. Chem. 4, 881 (1965). 25. Thoma, R. E., Hebert, G. M., Insley, H., and Weaver, C. F., Inorg. Chem. 2_, 1005 (1963). 26. Thoma, R. E., Insley, H., and Hebert, G. M., Inorg. Chem. 5, 1222 (1966). 27. Fleet, S.G., Zeit. Krist. 121, 349 (1965). 28. Van Tassel, R., Min. Mag. 29, 850 (1954). 29. Caruba, R., Baumer, A., and Turco, G., Compt. Rend. 2TOD, 2741 (1970). 30. Thibaut, G., Caruba, R., and Turco, G., Compt. Rend 274D, 792 (1972). 31. Beattie, I. R., Trans. Faraday Soc. 50, 581 (1954). 32. Schairer, J. F. and Bow en, N. L., Am. J. Sci. 253, 681 (1955). 33. Richerson, D. R., and Hummel, F. A., J. Am. Ceram. Soc. 55, 269 (1972). 34. Taylor, W. H., Z. Krist. 74,-! (1930). 35. Knowles, C.R., Rinaldi, F. F., and Smith, J .V., Indian Mineralogist 6, 127 (1965). 36. Peacor, D. R., Zeit. Krist. 127, 213 (1968). 37. St. v. Naray-Szabo, Zeit. Krist. 99, 277 (1938). 38. Newnham, R. E., Am. Mineralogist 52, 1515 (1967). 39. Beger, R. M. and Buerger, M. J., Proc. Nat. Acad. Sci. U. S. 58, 853 (1967). 69 OFFICIAL DISTRIBUTION LIST NASA NASA-Lewis Research Center (40) Attn: Mr. A. C. Antoine (M. S. 309-1) 21000 Brookpark Road Cleveland, OH 44135 NASA-Lewis Research Center Attn: L. W. Schopen (M. S. 500-206) 21000 Brookpark Road Cleveland, OH 44135 NASA-Lewis Research Center Attn: N. T. Musial (M. S. 500-113) 21000 Brookpark Road Cleveland, OH 44135 NASA-Lewis Research Center (2) Attn: Library (M. S. 60-3) 21000 Brookpark Road Cleveland, OH 44135 National Aeronautics and Space Administration Scientific and Technical Information Facility Attn: Acquisitions Branch P. O. Box 33 College Park, MD 20740 (12 copies and 1 reproducible) Mr. Simon Manson, Code ES National Aeronautics and Space Administration Washington, DC 20546 Mr. Floyd Ford, Code 761. Z Goddard Space Flight Center National Aeronautics and Space Administration Greenbelt, MD 20771 Mr. Gerald Halpert, Code 761. Z Goddard Space Flight Center National Aeronautics and Space Administration Greenbelt, MD 20771 Mr. Thomas Hennigan, Code 761 Goddard Space Flight Center National Aeronautics and Space Administration Greenbelt, MD 20771 Dr. Louis Rosenblum, MS 302-1 Lewis Research Center National Aeronautics and Space Administration 21000 Brookpark Road Cleveland, OH 44135 Mr. Harvey Schwartz, MS 309-1 Lewis Research Center National Aeronautics and Space Administration 21000 Brookpark Road Cleveland, OH 44135 Dr. J. Stuart Fordyce, MS 309-1 Lewis Research Center National Aeronautics and Space Administration 21000 Brookpark Road Cleveland, OH 44135 Mr. Charles B. Graff, S&E-ASTR-EP George C. Marshall Space Flight Center National Aeronautics and Space Administration Huntsville, AL 35812 Mr. Hoyt McBryar, EPS Johnson Space Center National Aeronautics and Space Administration Houston, TX 77058 JPL Mr. Daniel Runkle, MS 198-220 Jet Propulsion Laboratory 4800 Oak Grove Drive Pasadena, CA 91103 Dr. R. Lutwack, MS 198-220 Jet Propulsion Laboratory 4800 Oak Grove Drive Pasadena, CA 91103 Mr. Aiji A. Uchiyama MS 198-220 Jet Propulsion Laboratory 4800 Oak Grove Drive Pasadena, CA 91103 ARMY Harry Diamond Laboratories Room 300, Building 92 Connecticut Ave. & Van Ness St., N. W. Washington, DC 20438 U. S. Army Electronics Command Attn: AMSEL-TL-P Port Monmouth, NJ 07703 Commanding Officer U. S. Army Mobility Equip. Research & Development Center Electrotechnology Department Electrochemical Division Attn: SMEFB-EE Fort Belvoir, VA 22061 NAVY Director, Power Program, Code 473 Office of Naval Research Arlington, VA 22217 D2 Dr. George A. Neece Code 472 800 N. Quincy St. Office of Naval Research Arlington, VA 22217 Mr. S. Schuldiner, Code 6160 Naval Research Laboratory 4555 Overlook Avenue, S. W. Washington, DC 20375 Mr. J.H. Harrison, Code 2724 Naval Snip R&D Center Annapolis, MD 21402 Commanding Officer Naval Ammunition Depot (305, Mr. D. G. Miley) Crane, Indiana 47522 Mr. Phillip B. Cole, Code 232 Naval Ordnance Laboratory Silver Spring, MD 20910 Mr. Albert Himy, 6157D Naval Ship Engineering Center Center Bldg. , Prince Georges Center Hyattsville, MD 20782 Dr. H.E. RuskLe, NISC-4321 4301 Suitland Road Suitland, MD 20390 AIR FORCE AFAPL/ POE -1 / D. R. Warnock Wright-Patterson AFB, OH 45433 Air Force Aero Propulsion Laboratory POE-l/W. S. Bishop WPAFB, OH 45433 Mr. Edward Raskind, LCC, Wing F U. S. AF Cambridge Research Laboratory L. G. Hanscom Field Bedford, MA 01731 Rome Air Development Center Attn: TUGG/ F. J. Mollura Griffiss AFB, NY 13441 SAMSO/ DYAE P.O. Box 92960 Worldway Postal Center Los Angeles, CA 90009 OTHER GOVERNMENT ORGANIZATIONS Dr. Leonard Topper NSF, RANN Washington, DC 20550 PRIVATE ORGANIZATIONS Dr. E.A. Heintz Technical Department Airco Speer Carbon-Graphite P. O. Box 828 Niagara Falls, NY 14302 Dr. R. T. Foley Chemistry Department American University Massachusetts & Nebraska Aves., N. W. Washington, DC 20016 Mr. R. A. Knight Research Division AMF Inc. 689 Hope Street Stamford, CT 06907 Dr. H. Shalit ARCO Chemical Co. Division of Atlantic Richfield Co. 500 South Ridgeway Avenue Glenolden, PA 19036 Dr. James D. Birkett Arthur D. Little, Inc. Acorn Park Cambridge, MA 02140 Dr. H. L. Recht Atomics International Division International Corp. P. O. Box 309 Canoga Park, CA 91304 Mr. R. F. Fogle, GA 28 North American Rockwell Autonetics Division, NAR P.O. Box 4192 Anaheim, CA 92803 Mr. J. E. Clifford, Technical Representative Electrochemical Eng. Technical Division Battelle Columbus Laboratories 505 King Avenue Columbus, OH 43201 Mr. D. O. Feder Bell Telephone Laboratories, Murray Hill, NJ 07974 Dr. Carl Berger 13401 Kootenay Drive Santa Ana, CA 92705 Mr. Sidney Gross MS S+-37 The Boeing Company P. O. Box 3999 Seattle, WA 98124 Inc. D3 Professor T. P. Dirkse Calvin College 3175 Burton Street, S. E. Grand Rapids, MI 49506 Professor Ernest Yeager Department of Chemistry Case Western Reserve Univ. Cleveland, OH 44106 Mr. C. E. Thomas Chrysler Corporation Space Division Dept. 2730 P. O. Box 29200 New Orleans, LA 70189 Mr. E. P. Broglio Eagle-Picher Industries, Inc. P. O. Box 47, Couples Dept. Joplin, MO 64801 Dr. J. M. Williams Experimental Station, Bldg. 304 Engineering Technology Lab. E. I. du Pont de Nemours & Co. Wilmington, Delaware 19898 Mr. R. P. Mikkelson Electrical Systems Dept. 623-2 General Dynamics/ Convair Aerospace Division P. O. Box 80847 San Diego, CA 92138 Xerox Corporation Electro-Optical Systems 300 North Halstead Street Pasadena, CA 91107 Mr. Martin Klein Energy Research Corporation 15 Durant Avenue Bethel, CT 06801 Dr. J. G. Cohn Engelhard Industries Menlo Park Edison, NJ 08817 Mr. L. Berkowitz Government Research Lab. Esso Research and Engineering Co. P. O. Box 8 Linden, NJ 07036 Dr. Arthur Fleischer 466 South Center Street Orange, NJ 07050 The Garrett Corporation Suite 515, Cafritz Building 1625 Eye Street, N. W. Washington, DC 20006 Dr. J. B. Budi, Jr. Research and Development Center General Electric Company Bldg. K-l, Room 4A28 P. O. Box 8 Schenectady, NY 12301 Mr. Kenneth Hanson General Electric Company Valley Forge Space Technology Center P. O. Box 8555 Philadelphia, PA 19101 Mr. J. A. Keralla Delco Remy Division General Motors Corporation 2401 Columbus Avenue Anderson, IN 46011 Dr. John McCallum President Invention Talents, Incorporated 1149 Chesepeake Avenue Columbus, OH 43212 Mr. L. G. Nuttall General Electric Company 930 Western Avenue, 274A4 Lynn, MA 01910 Mr. F. T. O' Brien Direct Energy Conversion Programs General Electric Company 930 Western Avenue Lynn, MA 01910 Dr. E. L. Simons Environmental Protection Operation - Building 36 General Electric Company Schenectady, NY 12345 Dr. G. Goodman Globe-Union, Inc. P. O. Box 591 Milwaukee, WI 53201 Dr. J. E. Oxley Dr. B. B. Owens Gould Inc., Gould Lab. P. O. Box 3140 St. Paul, Minnesota 55165 Grumman Aerospace Corp. S. J. Gaston, Plant 35, Dept. 553 Bethpage, Long Island New York 11714 Gulton Battery Corporation 212 Durham Ave., Middlesex County Metuchen, NJ 08840 Dr. P. L. Howard Millington, MD 21651 D4 Dr. M. E. Ellion, Manager Propulsion & Power Systems Lab. Building 366, MS 524 Hughes Aircraft Company El Segundo, CA 90245 Mr. R. Hamilton Institute for Defense Analyses 400 Army-Navy Drive Arlington, VA 22202 Mr. James R. Hunt International Nickel Company 1000-16th Street, N. W. Washington, DC 20036 Dr. A. Moos Leesona Corporation Warwick, RI 02887 Dr. R. A. Wynveen, President Life Systems, Inc. 23715 Mercantile Road Cleveland, OH 44122 Mr. "Robert E. Corbett Department 62-25, Building 151 / 1 Lockheed Aircraft Corporation P. O. Box 504 Sunnyvale, CA 94088 Mr. S. J. Angelovich Chief Engineer Mallory Battery Company South Broadway Tarrytown, NY 10591 Mr. A. D. Tonelli Dept. A3-833, MS 22-2 McDonnell Douglas Astronautics Co. 5301 Bolsa Avenue Huntington Beach, CA 92647 Dr. Robert C. Shair 4921 Sanayer Dr. Hollywood, FL 33021 Rocketdyne Division North American Rockwell Corp. Attn: Library 6633 Canoga Avenue Canoga Park, CA 91304 National Center for Energy Management & Power 113 Towne Building University of Pennsylvania Phil., PA 19104 Dr. C. Bocciarelli 112 E. 2nd Street Moorestown, NJ 08057 Mr. D. C. Briggs WDL Division Philco-Ford Corporation 3939 Fabian Way Palo Alto, CA 94303 Dr. Per Bro P. R. Mallory & Company, Inc. Norlhwest Industrial Park Third Avenue Burlington, MA 01801 P. R. Mallory & Company, Inc. Library P. O. Box 706 Indianapolis, IN 46206 Mr. V. D'Agostino RAI Research Corporation 225 Marcus Blvd. Hauppauge, L. I., NY 11787 Southwest Research Institute Attn: Library P.O. Drawer 28510 San Antonio, TX 78284 Library Standord Research Institute 333 Ravenswood Avenue Menlo Park, CA 94025 Dr. W. R. Scott (Ml-1208) TRW Systems, Inc. One Space Park Redondo Beach, CA 90278 Dr. Herbert P. (R-l/ 2094) TRW Systems, Inc. One Space Park Redondo Beach, CA Silverman 90278 Union Carbide Corporation Battery Products Division Development Laboratory Library P. O. Box 6056 Cleveland, OH 44101 Dr. Robert Powers Consumer Products Division Union Carbide Corporation P.O. Box 6116 Cleveland, OH 44010 United Aircraft Corporation Attn: Library 400 Main Street East Hartford, CT 06108 Power Information Center University City Science Institute 3401 Market Street, Room 2210 Philadelphia, PA 19104 Dr. Frederick Morse Department of Mechanical Engineering University of Maryland College Park, MD 20742 D5 Yardney Electric Corporation Power Sources Division 3850 Olive Street Denver, CO 80207 Yardney Electric Division 82 Mechanic Street Pawcatuck, CT 02891 Dr. Eugene Y. Weissman, Director Inorganic-Electrolytic R&D v BASF Wyandotte Corporation Wyandotte, MI 48192 Dr. Charles Levine Dow Chemical U. S. A. Walnut Creek Research Center 2800 Mitchell Drive Walnut Creek, CA 94598 Prof. Donald M. Smyth Materials Research Center Lehigh University Bethlehem, PA 18015 Prof. John W. Patterson Dept. of Metallurgy Iowa State University Ames, IA 50010 Prof. Rustum Roy Materials Science Dept. Pennsylvania State University University Park, PA 16802 Prof. JohnH. Kennedy Univ. of Calif, at Santa Barbara Santa Barbara, CA 93106 Dr. Paul Jorgensen Stanford Research Institute Menlo Park, CA 94025 Dr. L. Topper Div. of Advanced Technology Applications National Science Foundation Washington, DC 20550 Mr. L. R. Rothrock Union Carbide Corp. 8888 Balboa Ave. San Diego, CA 92123 Prof. K. E. Cox Dept. of Chem. & Nucl. Eng. Tech. Appl. Center University of New Mexico Albuquerque, N. M. 87131 Dr. Robert A. Huggins Dept. of Materials Science and Engineering Stanford University Stanford, CA 94305 Dr. M. Stanley Whittingham Esso Research and Engineering Co. Linden, NJ 07036 Dr. Neill T. Weber Ford Motor Co. Research Lab. Dearborn, MI 48121 Prof. James I. Mueller Ceramic Engineering Div. University of Washington Seattle, WA 98195 Dr. Douglas O. Raleigh Nortii American Rockwell Science Center Thousand Oaks, CA 91360 Dr. R.H. Doremus Rensselaer Polytechnic Institute Materials Division Troy, NY 12181 Dr. Robert S. Roth National Bureau of Standards U. S. Department of Commerce Washington, DC 20234 Prof. Alexander F. Wells Dept. of Chemistry University of Connecticut Storrs, CT 06268 Dr. Elton J. Cairns » Electrochemistry Dept. General Motors Research Lab. 12 Mile and Mound Roads Warren, Michigan 48090 Dr. John B. Goodenough Massachusetts Institute of Technology Lincoln Laboratory Lexington, Mass. 02173

Ionic conductivity of lanthanum fluoride

Abstract

Extracted data

Ionic conductivity of lanthanum fluoride

Abstract

Extracted data

Related items

Related items